AI/ML Engineer - RAG Pipelines & Vector Search Expert

VARSHITH GADDAM

Hyderabad, India +91-950******* *************@*****.*** Portfolio GitHub LinkedIn SUMMARY

AI/ML Engineer (B.Tech 2026, CGPA 8.3) with 20+ end-to-end production AI projects. IEEE-published researcher

(ICSSAS 2026) and hackathon winner. Specialises in RAG pipelines, multi-agent orchestration (LangGraph), and vector search (Qdrant, Pinecone, FAISS). 365-day coding streak, 60+ open-source repositories. ACHIEVEMENTS

• IEEE Publication: Accepted author at ICSSAS 2026 (IEEE Xplore) – oral presentation on novel computer vision research.

• Hackathon Winner: 1st Place, regional technical hackathon – built and shipped a functional AI system within 24 hours.

• Coding Streak: 365-day LeetCode streak, 380+ DSA problems solved; 60+ public GitHub repositories.

• Certifications: Generative AI Essentials (Microsoft) Python with ML (Coursera) Python (University of Moratuwa). TECHNICAL SKILLS

GenAI and LLMs: LangChain, LangGraph, RAG Pipelines, Prompt Engineering, Function Calling, Guardrails, Streaming APIs

Machine Learning: CNNs, UNet, Regression, Classification, Forecasting, Scikit-learn, Pandas, NumPy, OpenCV Vector DBs and Retrieval: Qdrant, Pinecone, FAISS, Semantic Search, Re-ranking, Transformer-based Embedding Models

Engineering and DevOps: Python (Advanced), FastAPI, Docker, REST APIs, SQL, Git, CI/CD Cloud: GCP, AWS (S3, Lambda)

EXPERIENCE

AI/ML Engineer Intern – US-Based Client Project May 2025 – July 2025

• Built a financial document intelligence system using GenAI to extract structured insights from SEC filings at scale.

• Designed and optimised a full RAG pipeline – PDF chunking, embedding, vector retrieval, and re-ranking – cutting manual audit time by 40%.

• Engineered prompt structures for structured JSON output and function calling, improving downstream data relia- bility.

• Conducted model evaluation across groundedness, hallucination rate, and latency; documented findings for global engineering teams.

AI Intern – Evoastra Ventures Pvt Ltd 2025

• Built predictive ML models using CNNs and Scikit-learn for automated visual inspection pipelines.

• Performed feature engineering and data cleaning, improving model stability and reliability by 15%. KEY PROJECTS

LegalMind AI – Multi-Agent Research System Python, LangGraph, Qdrant, FastAPI

• Architected a multi-agent system using LangGraph orchestrating specialised agents for statutory analysis, citation retrieval, and reasoning synthesis.

• Built a high-performance re-ranking retrieval layer using Qdrant, improving citation accuracy by 35% over baseline vector search.

• Delivered real-time reasoning traces via a streaming FastAPI backend, enabling interpretable and grounded legal AI outputs.

Indian Legal Intelligence System Python, Qdrant, LangChain

• Built a Document Q&A platform mapping criminal and civil scenarios to Indian statutory law with 90%+ precision.

• Designed hybrid retrieval combining dense vector search and keyword matching, cutting retrieval latency by 30%. Math Professor Agent LangChain, FastAPI, Docker

• Engineered an agentic copilot with intelligent tool routing to autonomously solve multi-step academic queries end- to-end.

• Deployed as a Dockerised FastAPI service, maintaining low latency under concurrent user load. EDUCATION AND PUBLICATION

Vignan Institute of Technology and Science Graduated April 2026 B.Tech in Artificial Intelligence and Machine Learning CGPA: 8.3/10 Government Polytechnic College 2020 – 2023

Diploma in Mechanical Engineering CGPA: 9.83/10

Publication: MSF-UNet: Multi-Scale Flow Guided UNet for Enhanced Semantic Change Detection – 4th International Conference on Self Sustainable AI Systems (ICSSAS 2026), IEEE Xplore, May 2026. Proposed a novel multi-scale flow guided UNet architecture for remote sensing imagery; accepted for oral presentation. 1

Contact this candidate