Data Scientist | AI/ML Engineer | CV | NLP | GeN AI | Agentic AI Specialist
πDelhi , DL , INDIA |πSanFrancisco , CA , USA | πΌ 4+ Years Experience | π Building Production-Grade AI/ML Systems
π¬ Data Scientist (AI/ML Engineer) skilled in building, deploying, and optimizing end-to-end Machine Learning , Deep Learning , Computer Vision , NLP , Generative AI and Agentic AI solutions at scale.
- π Handling 1PB+ large datasets and developing real-time data pipelines
- π Delivering production-grade AI/ML/DL/CV/NLP/GenAI/Agentic AI systems across cloud ( AWS,AZURE,GCP ) environments
- π€ Building GenAI LLM-based chatbots, vector search systems, and secure, scalable enterprise applications
- π§ API development & integration, automation, and data engineering workflows
- π Breaking down complex problems, optimizing model performance, and driving measurable business outcomes
- Experience across diverse AI/ML algorithms: LR, SVM, Decision Trees, Random Forest, XGBoost, and Deep Learning models (CNN, RNN, GANs, Transformers, YOLOv26)
- Expertise in Computer Vision, NLP, Text Analytics,Generative AI ,Agentic AI systems, and business value analysis
- Hands-on background in algorithm design, model evaluation, error analysis, and performance optimization
- Successfully handled petabyte-scale (1PB+) data in real-world production environments
- Deployed ML / DL / CV / NLP / GenAI /Agentic AI models into production in close collaboration with engineering teams
Frameworks: LangChain β’ LangGraph β’ LangFlow β’ CrewAI β’ PhiData (Agno) β’ OpenAI Agents SDK β’ Autogen β’ LlamaIndex β’ MCP (Client/Server) β’ LangSmith
Capabilities: Tool Calling β’ Memory Systems β’ Multi-Agent Workflows β’ Agent Orchestration β’ FastAPI Integrations
Expertise:
- LLM Architectures β’ Prompt Engineering β’ RAG (Retrieval Augmented Generation)
- Fine-Tuning (SFT, LoRA, QLoRA) β’ Model Optimization & Quantization (GGUF, 4-bit/8-bit)
- OpenAI GPT Models β’ Llama3 β’ Mixtral β’ Claude (Anthropic) β’ Vertex AI
Capabilities: Embedding Models (OpenAI, HF, BGE, E5) β’ Hybrid Search (BM25 + Vector) β’ Reranking (Cohere/BGE) β’ Chunking Strategies
Architectures: β’ Transformers β’ CNN β’ RNN β’ LSTM β’ GANs β’ YOLOv26 β’ R-CNN
Applications: β’ OCR β’ Object Detection β’ Classification β’ Segmentation
Supervised Learning:
Linear Regression β’ Logistic Regression β’ Decision Trees β’ Random Forest β’ SVM β’ Naive Bayes β’ k-NN β’ Gradient Boosting β’ XGBoost β’ LightGBM
Unsupervised Learning:
k-Means β’ Hierarchical Clustering β’ DBSCAN β’ PCA β’ t-SNE β’ Autoencoders β’ GMM
Other ML:
Reinforcement Learning β’ Time Series (ARIMA, SARIMA, Prophet) β’ Recommendation Systems
Core: S3 β’ EC2 β’ Lambda β’ IAM β’ CloudWatch
AI/ML: SageMaker β’ Bedrock β’ Kendra β’ Guardrails
Data: EMR β’ DynamoDB β’ Redshift β’ Glue β’ Athena β’ OpenSearch
Integration: API Gateway β’ SNS β’ SQS β’ ECS
Analytics: QuickSight β’ CloudFormation β’ Cognito
Tools: AWS Ecosystem β’ MLFlow β’ Docker & DockerHub β’ Jenkins β’ Git/GitHub β’ GitLab β’ CI/CD β’ Streamlit β’ Pytest β’ Jira
Concepts: High Level Design (HLD) β’ Low Level Design (LLD) β’ Scalability β’ CAP Theorem β’ Sharding β’ Caching β’ Load Balancing
Developed and integrated RESTful APIs using FastAPI and FlaskAPI
| Duration | Institute | Degree & Specialization | GPA / CGPA | Links |
|---|---|---|---|---|
| Aug 2022 - 2024 | Woolf University | Master of Science (MS) Computer Science: Artificial intelligence and Machine Learning |
4.0 / 4.0 GPA | College | Degree | Academic institution |
| Jul 2016 - 2020 | Guru Nanak Dev Engineering College | Bachelor of Technology (B.Tech) Information Technology |
7.34 / 10.0 CGPA | College | Degree |
| Section | Topics | Sub-Topics | Sub Topics In Details | Live Project Details (Use Cases), Tech Stack & Links |
|---|---|---|---|---|
| A | Agentic AI & Gen AI | 01. 02. 03. 04. 1. Agentic AI & MCP 2. Gen AI, RAG, LLM 3. AIOps, LLMops using AWS services 4. UI/UX β Streamlit, ReactJS |
1. Agentic AI & MCP β’ Agentic AI: Autonomous agents that perceive, decide, and act (e.g., RPA, trading bots). β’ MCP: End-to-end framework for orchestration, versioning, A/B testing, and feedback loops. 2. Gen AI, RAG, LLM β’ Gen AI: Models generating text/images/code via massive unsupervised learning. β’ RAG: Enhances LLMs by fetching external knowledge to reduce hallucinations. β’ LLM: Transformer-based models (GPT-4, Llama) fine-tuned for specific tasks. 3. AIOps, LLMops (AWS) β’ AIOps: AI for IT ops (anomaly detection, root-cause analysis). β’ LLMops: Managing LLMs at scale (inference costs, latency). β’ AWS: SageMaker (End-to-end), Lambda (Serverless), CloudWatch (Monitoring). 4. UI/UX β Streamlit, ReactJS β’ UI/UX: Focuses on userβcentric designβintuitive interfaces and smooth experiences. β’ Streamlit: Python library that turns data scripts into shareable web apps instantly (great for ML demos). β’ ReactJS: JavaScript framework for building complex, stateβmanaged frontβends with reusable components. |
β’Agentic AI & GeN AI with Cloud's πCloud (AWS, GCP, AZURE ) β’[Details]β’[Tech Stack] β’Agentic BI SaaS 1.Agentic BI NLP Querying Analytics (LangGraph,Mem0,FAISS,Groq,VectorDb,MCP,FastAPI,SQL,NLP)β’ [Details] β’ [Tech Stack] β’ [Live Demo] β’MCP & A2A 1.Weather Agent (MCP & Agent to Agent) β’[Details] β’[Tech Stack] β’[Live Demo] β’Agentic AI 1.Agentic AI Travel Planner Itinerary(Langchain)β’[Details]β’[Tech Stack]β’[Live Demo] 2.Agentic AI Trip Planner (CrewAI)β’ [Details] β’ [Tech Stack] β’ [Live Demo] 3.Enterprise Multi-AI Agent Systems(LangGraph,Langchain,LlamaIndex) β’[Details] β’[Tech Stack] β’[Live Demo] 4.Agentic RAG Anime Recommender System(ChromaDB,HuggingFace Transformers,LangChain) β’[Details] β’[Tech Stack] β’[Live Demo] β’RAG & LLM 1.Universal PDF RAG Chatbot β’ [Details] β’ [Tech Stack] β’ [Live Demo] 2.Medical RAG Chatbot (HuggingFace,Langchain,Llama3) β’[Details] β’[Tech Stack] β’[Live Demo] β’GeN AI 1.AI-Teaching-Assistant β’[Details] 2.DeepTutor-STUDY-BUDDY-AI (ChromaDB,HuggingFace Transformers,LangChain,GroqAPI,TavilyAPI-Web Search ) β’[Details] 3.AI Enterprise Systems ChatGPT β’ [Details] 4.GenAI Music Composer(LangGraph,Langchain,LlamaIndex) β’[Details] β’[Tech Stack] β’[Live Demo] β’LLMOps & AIOPs 1.Flipkart-Product-Recommender-RAG(Llama 3 ,AstraDB ,HuggingFace ) β’[Details]β’[Tech Stack]β’[Live Demo] 2.GenAI Music Composer(LangGraph,Langchain,LlamaIndex) β’[Details] β’[Tech Stack] β’[Live Demo] 3.Celebrity Recognition & QA AI System(OpenCV,Vision Transformers, Groq LLaMA-4 Vision) β’[Details] β’[Tech Stack] β’[Live Demo] β’AGENTIC/GENAI,LLMOps & AIOps 8 Details End to End WORK β’[Live Details] |
| B | Deep Learning | 01. 02. 03. 04. 1. Neural Networks 2. Computer Vision 3. NLP (Natural Language Processing) 4. Transformer |
1. Neural Networks β’ Layers of interconnected neurons (input, hidden, output). β’ Forward pass computes outputs; backward pass (backpropagation) updates weights using gradients. 2. Computer Vision β’ CNNs (Convolutional Neural Networks): Extract spatial hierarchies in images via convolutions & pooling. β’ Architectures: YOLO (realβtime object detection), ResNet (deep networks with skip connections to avoid vanishing gradients). 3. NLP (Natural Language Processing) β’ Tasks: text classification, sentiment analysis, namedβentity recognition. β’ Techniques: word embeddings (Word2Vec, GloVe), sequence modeling (RNNs, LSTMs). 4. Transformer β’ Selfβattention mechanism that weighs input token relevance dynamically. β’ Enables parallel processing, improving performance on long sequences. β’ Variants: BERT (bidirectional, masked language modeling), GPT (generative, autoregressive). |
β’NLP & Generative AI πAdvance NLP & Generative AI β’[Details]β’[Tech Stack] β’ Deep Learning & Computer Vision-Production Ready 1.GenAI & Computer Vision β’[Details]β’[Tech Stack] β’Neural-Network 1.Neural Network Powered Delivery Time Estimation β’[[Details] β’Computer Vision 1.Tesla-Autonomous-Car-Driving-Vision-YOLOv5-Object-Detection β’ [Details]β’ [Tech Stack] β’ [Live Demo] 2.Defence AI: Multi-Sensor System β’ [Details]β’ [Tech Stack] β’ [Live Demo] 3.AI Driven Hotel Invoice Processing Pipeline β’[Details] 4.Agri_Tech-AI-Powered-Vegetable-Classifier β’[Details] β’NLP 1.Twitter-NER-System β’[Details] 2.FlipItNews-NLP-Classifier β’[Details] 3.AI-Powered FullStack News Classifier β’[Details]β’[Tech Stack] β’[Live Demo] 4.BERT embeddings with traditional NLP features β’[Details] β’Transformer 1.Fine-tuning Transformer Models Using PEFT (Parameter-Efficient Fine-Tuning) Techniques β’[[Details] |
| C | Machine Learning | 01. 02. 03. 04. 1. Maths for ML (Probability, Stats, Algebra, Calculus) 2. ML Types (Supervised, Unsupervised, RL, Time Series) 3. MLOps + FastAPI + Docker + AWS services |
1. Maths for ML β’ Probability: Distributions (Gaussian, Bernoulli), Bayes theorem for probabilistic models. β’ Statistics: Hypothesis testing, confidence intervals, regression analysis. β’ Linear Algebra: Matrix operations, eigenβdecomposition for PCA/dimensionality reduction. β’ Calculus: Gradient descent (optimization), chain rule for backpropagation. 2. Machine Learning types β’ Supervised Learning: Labeled data; algorithmsβlinear regression, SVM, random forests. β’ Unsupervised Learning: Unlabeled data; clustering (kβmeans), anomaly detection, PCA. β’ Reinforcement Learning: Agent learns via rewards/penalties; Qβlearning, Deep QβNetworks (DQN). β’ Time Series & Recommendation: ARIMA forecasting, LSTM for sequences; collaborative filtering for recommendations. 3. MLOps + FastAPI + Docker + AWS β’ MLOps: ML DevOpsβpipeline automation (CI/CD), model monitoring, reproducibility. β’ FastAPI: Highβperformance Python web framework for building REST APIs (serving ML models). β’ Docker: Containerization packages code + dependencies for consistent environments. β’ AWS deployment: EC2 (VMs), Lambda (serverless), SageMaker (managed ML services). |
β’MLOps Production-Ready-MLOps-Pipelines β’[Details] β’Time Series & Forecasting 1.AdEase AI Forecasting Engine β’[Details]β’[Tech Stack]β’[Live Demo] β’Recommendation-System 2.ZeeMovies Movie Recommendation Systemβ’[Details]β’[Tech Stack]β’[Live Demo] β’Manual Clustering(Unsupervised Clustering -K-means,Hierarchical Clustering) 3.EdTech Learner Clustering Analysisβ’[Details]β’[Tech Stack]β’[Live Demo] β’Ensemble Learning-Bagging & Boosting,KNN Imputation of Missing Values,Random Forest,XGBoost,Working with an imbalanced dataset 4.OLA Driver Churn Prediction β’[Details] β’[Tech Stack] β’[Live Demo] β’Feature Engineering,Logistic Regression, Precision Vs Recall Tradeoff 5.LoanTap-Credit-Risk-Analysis β’[Details] β’[Tech Stack] β’[Live Demo] β’Exploratory Data Analysis,Linear Regression,Statsmodels 6.Jamboree Education-Linear Regression, β’[Details] β’[Tech Stack] β’[Live Demo] β’Feature Creation,Relationship between Features,Column Normalization/Column Standardization,Handling categorical values,Missing values-Outlier treatment/Types of outliers 7.Delhivery Logistics-Feature Engineering β’[Details] β’[Tech Stack] β’[Live Demo] β’Bi-Variate Analysis,2-sample t-test: testing for difference across populations,ANNOVA,Chi-square 8.Yulu Bike-Hypothesis Testing β’[Details] β’[Tech Stack] β’[Live Demo] |
| D | Data Analyst | 01. 02. 03. 04. 1.Python & Libraries (NumPy, Pandas, EDA) 2.SQL 3.Statistics & Probability 4.Dashboard tools (Qlik Sense,Power BI,Tableau,Excel) |
1. Python & libraries β’ NumPy: Numerical operations on arrays. β’ Pandas: Data manipulation (DataFrame, cleaning, aggregation). β’ Matplotlib/Seaborn: Static & aesthetic visualizations. β’ SciPy: Scientific computing (optimization, stats). β’ EDA: Summarizing/visualizing data to find patterns or anomalies. 2. SQL β’ Querying: Joins, Window Functions, Aggregate functions. β’ Manipulation: DDL, DML, Indexing, and Optimization. 3. Probability & Statistics β’ Stats: Mean, Median, Mode, Standard Deviation, Hypothesis Testing. β’ Probability: Bayes Theorem, Distributions (Normal, Binomial), A/B Testing. 4. Dashboard tools β’ Power BI: Microsoftβs business analytics (dragβandβdrop reports, DAX). β’ Tableau: Interactive visualizations & dashboards. β’ Qlik Sense: Associative analytics for data discovery. β’ Excel: PivotTables, Power Query for basic analytics. |
β’Univariate & Bivariate & For continuous variable(s):Distplot,countplot,histogram for univariate analysis & For categorical variable(s):Boxplot & For correlation: Heatmaps,Pairplots 1.Walmart-Confidence Interval and CLT β’[Details] β’[Tech Stack] β’[Live Demo] β’EDA,correlations,outlier detection,segmentation,and 3D visualizations.using Python,Streamlit,Plotly,Pandas,NumPy,Seaborn & Matplotlib, Probability& Statistics. 2.Aerofit-Descriptive Statistics & Probability β’[Details] β’[Tech Stack] β’[Live Demo] β’Business-intelligence,EDA,using Python,Streamlit,Plotly,Pandas,NumPy,Seaborn & Matplotlib,content-strategy,Probability& Statistics. 3.Netflix-Data Exploration and Visualisation β’[Details] β’[Tech Stack] β’[Live Demo] β’SQL,DuckDB,comprehensive SQL-based insights,and dynamic Plotly visualizations for exploring sales trends,geography, logistics,and customer behavior. 4.Target Brazil E-Commerce Analytics Dashboard β’[Details] β’[Tech Stack] β’[Live Demo] |
| E | Data Engineering | 01. 02. 03. 04. 1. Big Data (Spark, Hadoop, Airflow, Kafka, ETL) 2. Data Warehouse & Databases (SQL, NoSQL, Snowflake) 3. AWS Services |
1. Big Data β’ PySpark: Python API for Spark; handles largeβscale data processing. β’ Apache Spark: Inβmemory computation engine for fast analytics (RDDs, DataFrames). β’ Hadoop: HDFS (storage) & MapReduce (batch processing). β’ Hive: SQLβlike queries over Hadoop data (data warehousing). β’ Airflow: Workflow orchestration (DAGs) for scheduling ETL jobs. 2. Data Warehouse & Databases β’ Data Warehouse: Optimized for readβheavy analytics (schemaβonβread, OLAP). Examples: Amazon Redshift, Snowflake. β’ NoSQL: Schemaβflexible databasesβdocument (MongoDB), wideβcolumn (Cassandra). β’ SQL Database: Relational DBs (MySQL, PostgreSQL) for structured queries & transactions. 3. AWS services β’ S3: Object storage for raw data lakes. β’ Glue: Serverless ETL service for data preparation. β’ Redshift: Fully managed data warehouse for analytics. β’ EMR: Managed Hadoop/Spark cluster for bigβdata processing. |
β’Azure Data Engineer 1.Azure Data Engineering Basic To Advanceβ’[Details] β’GCP Data Engineer 1.GCP Data Engineeringβ’[Details] β’AWS Data Engineer 1.Aws Data Engineeringβ’[Details] 2.AWS Services For Data Engineering With Projects β’[Details] β’Data Engineer Teck-Stack using AWS β’[Tech Stack] β’Data Pipeline using Kafka 1.Realtime Telecom Data Pipeline Kafkaβ’[[Details]β’[Tech Stack]β’[Live Demo] β’Data Pipeline Airflow-Kafka-Spark-Cassandra-Docker 2.Realtime Data Pipeline-Airflow-Kafka-Spark-Cassandra-Dockerβ’[[Details]β’[Tech Stack]β’[Live Demo] |
| F | Machine Learning & Data Engineering System Design | 01. 02. 03. 04. 1. High Level (HLD) & Low Level Design (LLD) 2. Scalability & Reliability (CAP, Load Balancing) 3. Distributed Systems & Microservices 4. Database Design (Sharding, Caching) |
1. High Level (HLD) & Low Level Design (LLD) β’ Every robust ML system (like a Recommendation Engine or Fraud Detection system) starts here. β’ Need HLD to map out how data flows from ingestion to training to inference, and LLD to design specific APIs. 2. Scalability & Reliability (CAP, Load Balancing) β’ Data Engineering deals with massive scale (Petabytes). β’ Understand how to scale horizontally (adding more servers) and ensure reliability when thousands of users hit your model. 3. Distributed Systems & Microservices β’ Big data tools (Spark, Kafka) are distributed systems. β’ Modern ML apps are built as microservices (e.g., separate services for data, model, UI) vs giant apps. 4. Database Design (Sharding, Caching) β’ Sharding: Essential when data is too big for one database (common in Data Eng). β’ Caching: Essential for low-latency ML inference (e.g., storing pre-calculated features in Redis). |
β’Q&A Ranking System (Quora/Reddit/Facebook) and E-commerce Promotion Forecasting (Amazon/Flipkart) 1.Machine Learning Model for Q&A Ranking β’[Details] β’Airbnb Home Value Prediction - End-to-End ML System Design 2.Airbnb Home Value Prediction β’[Details] β’A complete machine learning system that predicts the next app a user will open on their iPhone with 90% accuracy and <100ms latency. β’The Real-Time Fraud Analytics System is designed to detect fraudulent transactions in real-time for fintech applications.The system processes up to 10,000 transactions per second with sub-100ms latency. 3.Real-Time Fraud Analytics System β’[Details] β’A complete machine learning system that predicts the next app a user will open on their iPhone with 90% accuracy and <100ms latency. 4.AI Powered Next App Prediction β’[Details] β’Real-Time Data Streaming-Ingests and processes live stock data using Apache Kafka & Machine Learning Predictions-Uses an LSTM (Long Short-Term Memory) neural network to predict future stock prices in real-time. 5.Real-Time Stock Market Analysis System β’[Details] β’The Airline Ticket Shopping System is a comprehensive, production-ready ML platform built on AWS that enables airlines, travel agencies, and market. analysts to optimize pricing strategies, forecast demand, and provide personalized recommendations in real-time.Open-source ML community (XGBoost, scikit-learn, PySpark). 6.Airline ML Dynamic Pricing System β’[Details] β’Real-Time ETA Prediction System End-to-end ML system design for accurate food delivery time estimation using AWS services, advanced feature engineering, and gradient boosting models. 7.Food Delivery Order Real Time ETA ML Prediction System β’[Details] β’A comprehensive System Design and Prototype for a scalable,AI-driven photo organization platform similar to Google Photos.The complete Machine Learning System Design document.Includes architecture diagrams,component breakdown (Lambda, Rekognition, OpenSearch),and data flow strategies. 8.AI-Powered Photo Organizer β’[Details] β’Audio-Recognition-System Design ,The heart of the Shazam app is its ability to recognize songs through a process called audio fingerprinting.Similar to human fingerprints, each piece of music has a unique identifier that Shazam uses to identify songs from short audio snippets. 9.Audio-Recognition-System β’[Details] |
| G | Competitive Programming | 1. Algorithms 2. Data Structures |
1. Problemβsolving frameworks β’ Understand constraints, optimize time/space complexity (Big O). β’ Techniques: Two Pointers, Sliding Window, Bit Manipulation, Recursion. 2. Algorithms β’ Core: Sorting (Merge/Quick), Searching (Binary Search). β’ Advanced: Dynamic Programming (DP), Greedy, Backtracking, Graph Algorithms (BFS/DFS, Dijkstra). 3. Data Structures β’ Linear: Arrays, Linked Lists, Stacks, Queues, Hash Maps. β’ Trees & Graphs: Binary Trees, BST, Heaps, Tries, Disjoint Sets. |
β’Algorithms-and-Data-Structures Note -(Including all coding plateform's 5000+ Problems/Questions solved )β’ [Details]        |
A next-generation Agentic AI system that uses the Model Context Protocol (MCP) to standardize tool usage, enabling Llama 3 to fetch real-time global weather data with sub-second latency and zero hallucinations.
- Tech: MCP Server/Client β’ Llama 3 (Groq LPU) β’ LangChain β’ Streamlit β’ Open-Meteo β’ WebRTC
- Features: Universal tool protocol implementation, multi-modal voice interaction, sub-second reasoning, robust NLP city extraction, and interactive architectural visualization.
A production-grade Agentic Generative AI application that leverages autonomous AI agents to research, plan, and generate hyper-personalized travel itineraries with zero hallucinations.
- Tech: Groq LPU (Llama-3) β’ LangChain β’ Streamlit β’ Docker β’ Kubernetes β’ ELK Stack (Elasticsearch, Logstash, Kibana)
- Features: Autonomous reasoning workflows, sub-second itinerary generation, interactive premium UI, full LLMOps monitoring pipeline, and cloud-native microservices architecture.
Multi-agent AI system for intelligent travel planning using CrewAI framework
- Tech: CrewAI β’ LangChain β’ Multi-Agent Orchestration
- Features: Automated itinerary generation, budget optimization, personalized recommendations
A production-grade Agentic Retrieval-Augmented Generation (RAG) system that leverages semantic search and LLM reasoning to provide context-aware anime discovery with sub-second latency and zero hallucinations.
- Tech: Groq LPU (Llama 3.1) β’ LangChain β’ ChromaDB β’ HuggingFace Embeddings β’ Streamlit β’ Docker β’ Kubernetes (GKE)
- Features: Semantic plot & vibe discovery, agentic reasoning layers, interactive multi-tab premium UI, cloud-native containerization, and advanced LLMOps observability.
Advanced LangGraph-driven multi-agent ecosystem designed for high-speed intelligent reasoning and real-time orchestrated web research.
- Tech: LangGraph β’ Groq (Llama 3.1) β’ Tavily API β’ FastAPI β’ Docker β’ Jenkins β’ AWS
- Features: Cyclic agentic workflows β’ Near-zero latency inference β’ Automated DevSecOps/LLMOps/AIOps pipeline with SonarQube quality gates
Advanced RAG-based chatbot for PDF document Q&A
- Tech: LangChain β’ FAISS β’ OpenAI Embeddings β’ RAG
- Features: Multi-document support, semantic search, context-aware responses
A production-grade AI health assistant that delivers accurate, evidence-backed answers from medical encyclopedias using Retrieval-Augmented Generation (RAG) to eliminate hallucinations.
- Tech: LangChain β’ Llama 3 (HuggingFace) β’ FAISS β’ Streamlit β’ Docker β’ Jenkins β’ AWS App Runner β’ Aqua Trivy
- Features: Source-cited medical answers, sub-second vector retrieval, hallucination-free context injection, interactive system architecture visualization, and automated CI/CD deployment pipeline.
AI-powered recommendation engine using Retrieval-Augmented Generation for intelligent product discovery
- Tech: LangChain β’ Groq (Llama 3) β’ AstraDB β’ HuggingFace β’ Streamlit
- Features: Semantic search, review sentiment analysis, context-aware recommendations, real-time RAG pipeline
Production-grade generative AI orchestration studio that transforms natural language prompts into high-fidelity musical compositions using sub-second LLM inference.
- Tech: Groq LPU (Llama 3.1) β’ LangChain β’ Music21 β’ Synthesizer β’ Docker β’ GKE (Kubernetes)
- Features: Real-time melody & harmony orchestration, automated music theory validation, multi-tab operational monitoring, and professional WAV/MIDI export capabilities.
A state-of-the-art multimodal AI application combining computer vision for real-time identity detection with Large Language Models for context-aware celebrity Q&A.
- Tech: Python β’ Streamlit β’ OpenCV β’ Vision Transformer (ViT) β’ Groq LLaMA-4 Vision β’ Docker β’ Kubernetes (GKE)
- Features: Real-time biometric recognition β’ 128-d vector embeddings β’ Multimodal contextual reasoning β’ Low-latency inference β’ Microservices architecture
π Agentic BI SaaS: Natural Language Analytics Platform (LangGraph,Mem0,FAISS, Groq, VectorDb,MCP,FastAPI,SQL,NLP)
A production-grade Business Intelligence platform that replaces static dashboards with conversational analytics. Uses a 6-agent cognitive swarm orchestrated by LangGraph to transform natural language questions into real-time SQL queries, visualizations, and insights with enterprise-level governance.
- Tech: LangGraph β’ LangChain β’ Groq (Llama 3.3) β’ HuggingFace β’ FAISS β’ FuzzyWuzzy β’ FastAPI β’ Streamlit β’ Plotly β’ SQLAlchemy β’ Mem0 β’ MCP β’ Docker β’ Git LFS
- Features: Self-healing SQL generation with reflexion loops, typo-resilient query processing, FAISS-powered semantic search, real-time agent transparency logs, RBAC governance with PII detection, long-term user preference memory, and production-ready SaaS/On-Prem deployment architecture.
Real-time news classification using advanced NLP techniques
- Tech: BERT β’ Transformers β’ NLP β’ Classification
- Features: Multi-class news categorization, sentiment analysis
π¦ Twitter NER System
Named Entity Recognition system for social media text
- Tech: SpaCy β’ NER β’ NLP β’ Entity Extraction
- Features: Real-time entity detection, visualization, custom entity types
Real-time object detection for autonomous driving scenarios
- Tech: YOLOv5 β’ OpenCV β’ Computer Vision
- Features: Multi-object detection, real-time processing, bounding box visualization
Advanced surveillance system using YOLOv8
- Tech: YOLOv8 β’ Multi-sensor Fusion β’ Real-time Detection
- Features: Threat detection, multi-camera support, alert system
Advertising effectiveness forecasting using time series models
- Tech: ARIMA β’ SARIMA β’ Prophet β’ Time Series
- Features: Trend analysis, seasonality detection, future predictions
Student segmentation for personalized learning
- Tech: K-Means β’ DBSCAN β’ Clustering β’ Unsupervised Learning
- Features: Student profiling, learning pattern analysis, recommendations
Credit risk assessment and loan default prediction
- Tech: XGBoost β’ Random Forest β’ Classification β’ Risk Modeling
- Features: Credit scoring, risk stratification, feature importance analysis
Comprehensive logistics and supply chain analytics
- Tech: Pandas β’ Plotly β’ Data Visualization β’ Dashboard
- Features: Route optimization, delivery time prediction, KPI tracking
ML-based graduate school admission probability calculator
- Tech: Logistic Regression β’ Feature Engineering β’ Classification
- Features: Admission probability, university recommendations, profile analysis
Demand forecasting for bike-sharing services
- Tech: Time Series β’ Regression β’ Demand Forecasting
- Features: Hourly demand prediction, weather impact analysis, station optimization
Customer segmentation and product recommendation system
- Tech: Clustering β’ RFM Analysis β’ Customer Analytics
- Features: Customer profiling, product affinity, targeted marketing insights
Sales pattern analysis and revenue optimization
- Tech: EDA β’ Statistical Analysis β’ Visualization
- Features: Purchase behavior analysis, demographic insights, sales forecasting
Driver retention prediction using machine learning
- Tech: Gradient Boosting β’ Feature Engineering β’ Classification
- Features: Churn probability, retention strategies, driver profiling
Content analysis and recommendation insights
- Tech: NLP β’ Content Analysis β’ Recommendation Systems
- Features: Genre analysis, content trends, viewer preferences
E-commerce performance and customer behavior analysis
- Tech: SQL β’ Python β’ Business Intelligence
- Features: Sales metrics, customer lifetime value, product performance
Streaming data pipeline for telecom analytics
- Tech: Apache Kafka β’ Streaming β’ Real-time Processing
- Features: Live data ingestion, stream processing, real-time dashboards
End-to-end big data pipeline with orchestration
- Tech: Airflow β’ Kafka β’ Spark β’ Cassandra β’ Docker
- Features: Automated workflows, distributed processing, scalable architecture
- Developed cutting-edge AI models for computer vision applications
- Implemented deep learning solutions for medical image analysis
- Created real-time object detection systems with optimized performance
- Conducted workshops on Machine Learning and Deep Learning
- Mentored students in AI/ML projects and research
- Created educational content for Computer Vision courses
- π Deep Learning Specialization - Coursera
- π Machine Learning Engineering - Google Cloud
- π Computer Vision Nanodegree - Udacity
- ποΈ Best AI Project Award - University Hackathon 2024
class CurrentFocus:
def __init__(self):
self.learning = ["Agentic AI","MCP Server/Client","GenAI/RAG/LLM","NLP"]
self.learning = ["Advanced Computer Vision", "LLM Fine-tuning", "MLOps/AIOps/LLMOps","Machione Learing"]
self.building = ["AI-Powered Applications", "Real-time Systems"]
self.exploring = ["Generative AI", "Edge AI", "Federated Learning"]
self.open_to = ["Collaborations", "Open Source", "Job Opportunities"]
def get_status(self):
return "π Always learning, always building!"
me = CurrentFocus()
print(me.get_status())- Contribute to 151+ open-source AI projects
- Publish research papers on Computer Vision
- Build and deploy 75+ production-ready AI applications
- Mentor 50+ aspiring AI/ML engineers
- Master advanced LLM architectures and deployment
