Standing Rank
Rank Not Available
Developer Badges
Badges Pending
Skills & Technologies
52 skillsWork Experience
Generative AI Engineer
BharatPe (Fintech)
- Architected and deployed an LLM-powered agentic AI ticket deflection system on AWS Bedrock with RAG pipelines and FAISS vector search — reduced manual support load by 40% with 99% factual accuracy across 10,000+ daily AI interactions.
- Built scalable MLOps-ready backend serving 10,000+ daily LLM inference requests at 99.9% uptime; implemented multi-layer Redis caching reducing inference latency by 60%.
- Developed persona-driven chatbots with precision RAG retrieval, increasing CSAT scores by 25% by aligning customer feedback loops with LLM response tuning.
- Built a RAG pipeline with LangChain and Pinecone to automate compliance document processing — replaced legacy ETL with vector embeddings and semantic search, cutting manual data entry by 50%.
- Directed responsive React/TypeScript interfaces with performance tuning and layout optimization, improving user engagement by 30%.
AI/ML Intern (NLP & Data Science)
Kalpavruksh Technologies
- Developed and fine-tuned a Llama 3 customer support chatbot using Hugging Face; built end-to-end LangChain orchestration layer reducing FAQ response time by 50%.
- Optimized the model using RAG with LangChain, Hugging Face, and FAISS — boosted retrieval performance by 60%.
- Applied prompt engineering techniques to elevate response quality by 30%; established LLM evaluation protocols for accuracy and precision across diverse document formats.
- Built automated ETL pipeline to process unstructured PDF data into searchable vector embeddings for internal knowledge retrieval.
Data Analysis Intern
Cube Advertising
- Executed RFM (Recency, Frequency, Monetary) analysis on 50,000+ records — identified 12% growth opportunity in high-value customer segments.
- Built automated dashboards transforming raw marketing metrics into actionable insights; developed statistical segmentation models to improve targeting and ROI.
Projects
Enterprise RAG System with Multi-Agent Role-Based Intelligence
- Architected a role-aware LLM system ingesting multi-format data (PDF, JSON, CSV) with layered RAG pipeline (UI → FastAPI → Pinecone) supporting Tech, Product, and Compliance personas.
- Achieved 98% factual accuracy through high-precision semantic retrieval, real-time document re-ranking, and audit-ready source attribution.
Low-Code Agentic AI Pipeline Builder
- Built a visual node-based canvas enabling non-technical users to construct complex LLM sequences; engineered custom backend for DAG validation and cycle detection.
- Implemented type-inference linting system reducing pipeline configuration errors by 70%; added real-time collaborative building via Socket.io.
Generative Transformer from Scratch (GPT-2 Style)
- Built a complete transformer architecture from scratch using NumPy and PyTorch, resembling GPT-2 — covering attention mechanisms, positional encoding, and autoregressive decoding.
- Used pre-trained Hugging Face tokenizers and Rouge-L scoring for evaluation; leveraged Lightning for clean, scalable training code.
Fine-Tuned QA Model — DistilBERT on SQuAD
- Fine-tuned DistilBERT on the SQuAD dataset for extractive question answering; introduced custom feature modifications to enhance model performance.
Semantic Requirement-to-Code Validator
- Used Tree-sitter AST parsing to extract semantic code chunks; stored and compared against embedded requirements via LLM reasoning in ChromaDB.
- Automated ’Implementation Status Reports’ identifying missing features with 90%+ precision — significantly reducing manual QA cycles.
Leaderboard Standings
Leaderboard Position Pending
Global test scores, peer standing percentiles, and algorithm leaderboard ranks are updated dynamically.
Assessment Highlights
Assessments Not Completed
Coding evaluations, system assessment results, and conceptual score badges will appear here after taking a test.
AI Collaboration Score
AI Collaboration Score Pending
Developer coding behavior, assistant cooperation, and AI pair-programming indicators are evaluated during live coding sessions.
Role Compatibility Profile
Role Compatibility Analysis Pending
Custom matching reports, candidate role compatibility percentiles, and core engineer strength profiles are processed once conceptual code screenings are complete.
Achievements
JEE Merit
Top 1% ranker in Joint Entrance Examination (JEE) 2021; Ranked 126 in KEAM 2021.
IIM-A Finalist
National Finalist (Top 8 of 1,300+ teams) — FMCG Case Study, TRBS IIM Ahmedabad.
Case Competitions
Qualified Round 2 — Hospitality Industry Case Study, SARK TANK.
About Details
Professional Bio
Generative AI Engineer with production experience building LLM-powered systems using AWS Bedrock, LangChain, and agentic AI frameworks. Specialized in RAG pipeline design, vector database architecture, prompt engineering, and multi-agent systems.
B.Tech in Metallurgical Engineering
Indian Institute of Technology (BHU) Varanasi, India (2021 - 2023)