Projects

AI Trading Agent

Designed an AI trading agent using LangChain and Llama 3-Instruct to automate screening, ranking, and backtesting undervalued stocks, integrating reinforcement learning for adaptive investment decisions. Built ML/DL models using Python (TensorFlow, PyTorch, Decision Tree, Random Forest, XGBoost) on large YFinance and AlphaVantage datasets (60+ years data), achieving 92% prediction accuracy for investment strategy.

Tech Stack: LangChain, Llama 3, Python, TensorFlow, PyTorch, XGBoost, CUDA, LLM, TensorRT

Role: Research Assistant at NYU (Jan 2025 - Present)

Agentic AI Resume Intelligence System

Built an Agentic AI Resume Intelligence System (Llama 3-70B / GPT-4 + LangChain) to parse and match resumes to JDs, automating screening and boosting hiring accuracy by 90%. Engineered LangGraph + FastAPI multi-agent workflows integrating FAISS retrieval and Sentence-Transformer embeddings, improving match accuracy by 35% and reducing latency by 50%. Architected a cloud-native GCP pipeline (VMs, CI/CD, Pub/Sub, Docker, TensorRT) handling 500K+ resumes/day with continuous pipeline optimization improving sub-second latency by 80%. Achieved 87% LLM precision through automated A/B validation, iterative prompt optimization, and vector feedback fine-tuning, enhancing candidate matching accuracy and improving system reliability.

Tech Stack: Python, FastAPI, LangGraph, Llama 3, FAISS, Sentence Transformers, Angular, GCP, Docker, TensorRT

Role: AI Intern at STL Digital (Jun 2025 - Aug 2025)

CodeSensei – Autonomous Code Reviewer

Developed CodeSensei, an autonomous code-review agent using GPT-4 and AST parsing to detect logic flaws and propose optimized patches, reducing manual review effort by 55%. The system leverages advanced language models and code analysis techniques to provide intelligent code review suggestions, improving code quality and developer productivity.

Tech Stack: LangGraph, Claude/GPT-4, GitHub API, OpenAI Embeddings

Yuka Food & Cosmetic Scanner

Built an Agentic AI layer over Yuka using RAG, vector search, and GPT-4 to autonomously analyze ingredients and suggest safer alternatives, reducing manual lookups by 65%. The system uses advanced retrieval-augmented generation techniques to provide intelligent ingredient analysis and health recommendations.

Tech Stack: Python, LangChain, OpenAI GPT-4, FAISS, REST APIs

Professional Experience Projects

LTIMindtree Disney - Senior Software Engineer (Jul 2022 - Jul 2024)

• AWS Cloud Optimization: Redesigned AWS Cloud modules by containerizing microservices with Docker and optimizing ETL pipelines, achieving a 70% reduction in workflow complexity and 99.99% platform uptime.
• REST API Architecture: Resolved critical service bottlenecks by engineering modular REST API connectors and event-driven synchronization between Python orchestration modules and AWS S3, boosting data scalability by 40%.
• Automated Workflows: Automated region-specific audit and authorization workflows using Python schedulers and MySQL triggers, scaling to 2M+ users across 50+ regions and managing a 50B+ record database.
• Team Leadership: Led a 5-developer team to enhance the AWS S3 document retrieval system, reducing retrieval time by 70% through collaborative problem-solving, code reviews, and sprint planning.

Tech Stack: Python, AWS (S3, Lambda), MySQL, Docker, REST APIs, Redis, ETL Pipelines

LTIMindtree CitiBank - Senior Software Engineer (Jul 2021 - Jul 2022)

• GCP Systems Optimization: Resolved 500+ L3 issues in Citibank’s GCP systems using FAST APIs, Python automation, and SQL, improving performance and reliability across 50M+ daily transactions.
• Proof-of-Concepts: Architected React, Python, and SQL proof-of-concepts with BAs and product owners to resolve critical production bottlenecks and streamline enterprise workflows.

Tech Stack: Python, Java, FAST APIs, GCP (Pub/Sub, Cloud SQL, Load Balancer), Kubernetes, Microservices