VAIBHAV YADAV
Sector-**, Noida +91-707******* *********@*****.*** LinkedIn GitHub LeetCode SUMMARY
AI/ML engineering student with hands-on experience building and deploying end-to-end machine learning systems — from a production RAG application serving real clinical workflows to a Transformer implemented from the original research paper. Strong foundation in Python, PyTorch, LangChain, and full-stack ML deployment (FastAPI, MongoDB). 468+ LeetCode problems solved. Seeking an AI/ML Engineer internship to contribute to production-grade LLM and RAG systems. EDUCATION
B.Tech, Computer Science — JSS Academy of Technical Education, NoidaOct 2023 – Sept 2027 PROJECTS
AI Doctor Assistant FastAPI · Streamlit · LangChain · RAG · MongoDB · FAISS · Groq github.com/Vinsmoke-R/AI-Doctor-Assistant-
● Built a full-stack, cloud-deployed patient management system that eliminates shift-handover information loss for nursing teams by centralising all patient records, lab reports, and prescriptions in a single AI-queryable interface.
● Engineered an OCR pipeline using Tesseract to extract raw text from uploaded lab reports and prescriptions, then cleaned and structured the output using Groq Llama-3.3-70B before embedding with BAAI/bge-base-en-v1.5 into an in-session FAISS vector store.
● Designed a RAG pipeline where clinical queries trigger FAISS similarity search over report embeddings; the top-k chunks plus the patient's full profile are injected into the LLM prompt, grounding answers in actual document content rather than model memory.
● Implemented full CRUD patient record management backed by MongoDB Atlas, including auto-generated unique patient UIDs
(PAT-XXXXXXXX), base64 report storage, and persistent per-patient chat history across sessions.
● Deployed end-to-end on free-tier infrastructure — Streamlit Cloud (frontend) and Render (FastAPI backend) — with isolated error handling so a failed OCR extraction on one report never crashes the rest of the patient view. AI Stock Trading Agent LangGraph · LangChain · Groq · yfinance · Alpaca · Streamlit github.com/Vinsmoke-R/stock_trading_AI_Agent
● Built a LangGraph-orchestrated autonomous trading agent that combines technical analysis with real-time news sentiment to produce BUY / SELL / HOLD decisions and execute paper trades on Alpaca Markets.
● Implemented a multi-step agentic pipeline: fetch 1-year OHLCV price data (yfinance) compute RSI, EMA, VWAP, ATR signals
pull live financial headlines (feedparser) score news sentiment with Llama-3.3-70B pass all signals to the trading model for a final reasoned decision.
● Integrated Alpaca paper trading API to execute live buy/sell orders from the agent's output, enabling risk-free strategy validation in real market conditions.
● Built a Streamlit + Plotly dashboard surfacing live price charts, computed indicator overlays, sentiment scores, and agent reasoning in a single view.
Transformer Architecture from Scratch PyTorch · Deep Learning · NLP github.com/Vinsmoke-R/Transformer_from_scratch
● Reproduced the full Transformer architecture from the original 'Attention Is All You Need' paper in PyTorch, implementing every component without high-level attention APIs, as a deep-dive into the mechanics behind modern LLMs.
● Built multi-head self-attention, scaled dot-product attention, masked self-attention (decoder), positional encoding, residual connections, and layer normalization as clean, standalone PyTorch modules.
● Constructed a complete 6-layer encoder / 6-layer decoder pipeline for sequence-to-sequence modeling, faithfully following the paper's architecture without recurrence or convolution. TECHNICAL SKILLS
● Languages: Python, Java
● ML / DL: PyTorch, TensorFlow, scikit-learn, HuggingFace Transformers
● LLM / RAG / GenAI: LangChain, LangGraph, RAG pipelines, FAISS, Vector Databases, Groq API, OCR (Tesseract)
● Backend & Data: FastAPI, MongoDB / MongoDB Atlas, SQL, NumPy, Pandas, Matplotlib
● Tools & Platforms: Git, Streamlit, Plotly, Figma, Cursor ACHIEVEMENTS
● Solved 468+ problems on LeetCode, demonstrating strong data structures and algorithms fundamentals across arrays, trees, graphs, and dynamic programming.