Location // Etawah, Uttar Pradesh, India
Focus // Agentic AI, LLM Systems, Quant Finance, Scientific ML
Current_Mission // Building production-grade AI systems that translate complexity into actionable intelligence
Status // CS Student @ VIT Bhopal (AI & ML) • Founder @ Quant ML • Research Author × 3 Publications

Intelligent Research Assistant with Multi-Source Querying

Built an agentic research assistant that decomposes complex queries into subtasks and searches ArXiv, DuckDuckGo, and Wikipedia in parallel using LangGraph orchestration and multi-agent coordination.

Tech Stack: Python • LangChain • LangGraph • Multi-Agent Architecture • REST APIs • MIT License

→ Repository

10-Model Agentic Consensus System for Institutional Quant Finance

Architected a LangGraph-orchestrated 10-model AI Council with structured multi-agent consensus (≥99% threshold before execution). Includes FastAPI backend, Redis caching, PostgreSQL audit trail, and Prometheus/Grafana monitoring — FINRA-compliant, production-ready agentic workflow.

Highlights:

• 300× faster Heston model calibration via neural network surrogate; MSE < 0.001, inference < 50ms
• NIFTY50 pipeline: Sharpe Ratio 1.35 vs 0.72 buy-and-hold benchmark
• AI Investment Advisor RAG agent: +18.50% vs S&P 500, Sharpe 2.10
• Structured output pipelines with Pydantic schemas eliminating silent agent failures

Tech Stack: Python • LangGraph • FastAPI • Redis • PostgreSQL • FinBERT • RAG • Prometheus • Docker

→ Quant ML • → AI Advisor Repository

Production LLM Workflow for Multi-Channel Message Triage

Built a 6-class priority classifier agent (auto-send / agent-review / escalate) with structured outputs, adversarial security guardrails, explicit decision boundaries, and 14 automated tests passing. Mirrors real-world agentic workflow patterns end-to-end.

Tech Stack: Python • FastAPI • Claude API • PostgreSQL • LLM Reasoning

→ Repository

Full Retrieval-Augmented Generation Pipeline

End-to-end RAG pipeline over 500+ TF documentation pages: automated crawl → chunk → embed → vector-index → GPT-4 answer. Source-cited, syntax-highlighted, evaluatable outputs via Streamlit UI and CLI. Benchmarked across chunking strategies and embedding models.

Tech Stack: RAG • LangChain • GPT-4 • FAISS • Chroma • Supabase • Streamlit

→ Repository

Open-Source ML Monitoring & Pathology Detection Library

Open-source Python library auto-diagnosing 10+ ML training pathologies (overfitting, unstable LR, exploding gradients, early stopping issues) using transparent rule-based detectors with confidence scoring, severity levels, and actionable fix recommendations. Zero-config, fully offline, CI/CD ready.

Tech Stack: Python • Open Source • MIT License • CI/CD Integration

→ Repository

End-to-End Deep Learning for Automatic Sleep Classification

End-to-end deep learning pipeline for automatic sleep stage classification from polysomnography (PSG) signals. Classifies 30-second EEG/EOG/EMG epochs into 5 AASM sleep stages using knowledge distillation from Transformer models into lightweight 1D-ResNet for embedded-device inference.

Tech Stack: Python • PyTorch • CNNs • Transformers • TFLite • MIT License

→ Repository

NLP + LLM Browser Extension for Fraud Detection

Chrome extension + Python backend classifying URLs and financial offers using NLP and LLM-based reasoning. Demonstrates end-to-end AI product architecture from browser UI to backend LLM workflow.

Tech Stack: Python • FastAPI • NLP • LLM • Chrome Extension • MIT License

→ Repository

Predictive Intelligence for SpaceX Falcon 9 Landing Success

Engineered a complete ML pipeline achieving 85.19% accuracy predicting Falcon 9 first-stage landing success using XGBoost, SQL analytics, and Streamlit dashboards.

Tech Stack: Python • XGBoost • Scikit-Learn • SQL • Streamlit • Pandas • NumPy

→ Repository

Advanced Physics Simulation System

Computational system solving quantum mechanics problems with analytical and numerical Schrödinger-equation solvers. Accurate prediction of molecular absorption spectra and validation of foundational quantum-mechanical principles.

Tech Stack: Python • NumPy • SciPy • Matplotlib • Seaborn

→ Repository

• "PIGNet V2: Physics-Informed Graph Neural Networks for High-Throughput Crystalline Material Property Prediction" — ResearchGate, May 2026
• "Mechanistic Transparency of Neural Networks: A Four-Layer Framework" — ResearchGate, Jan 2026. 87.5% monosemantic neurons; ROME causal intervention; multi-step proofs linking representations to model behavior.
• "Liberating Justice: Fighting Judicial Waithood with AI" — ResearchGate, Dec 2025. RAG magistrate system with multi-step CoT and verifiable citations.

• Oracle Cloud Infrastructure 2025 Certified Generative AI Professional
• Google for Startups AI Fellow — Prompt to Prototype (Google × Scaler)
• IBM Generative AI: Elevate Your Data Science Career
• Columbia+: Prompt Engineering & Programming with OpenAI
• Google Cloud – Introduction to Large Language Models

• Harvard CS50 AI with Python
• IBM Data Science Professional Certificate
• WorldQuant Applied AI Lab — Deep Learning Fundamentals (2026)
• WorldQuant Applied AI: Computer Vision
• DeepLearning.AI NLP in TensorFlow
• Kaggle Intermediate Machine Learning
• Kaggle Intro to Machine Learning
• University of London – Machine Learning for All

• IBM Python for Data Science, AI & Development
• Google Crash Course on Python
• Cisco Python Essentials 1 & 2
• Infosys Springboard: Mastering Python

• IBM Databases and SQL for Data Science
• MongoDB Basics for Students
• Infosys Springboard: Hands-On Version Control with Git

• McKinsey Forward Program (Trainee)
• Google Analytics Certification
• Deloitte Data Analytics Job Simulation

Jan 2025 – Present

Building institutional-grade AI trading infrastructure targeting $1–5B AUM hedge funds. Architected multi-agent LangGraph systems, RAG pipelines, structured output frameworks, and quantitative finance models.

Jan 2026 – Apr 2026 · Remote

Authored institutional investment briefs via multi-step Python reasoning chains. Built data pipelines for market analytics; work validated by registered SEBI analysts.

Nov 2025 – Dec 2025 · Remote

Validated AI prototypes against real-world criteria using Gemini, Google AI Studio, and NotebookLM. Delivered a functional AI prototype at the Build the Future showcase.

Jul 2025 – Nov 2025

Shipped production Python/ML modules to multiple repos; collaborated with global maintainers to merge production-quality code.

VIT Bhopal × IIT Indore × CSIR-INDIA · 100+ participants

Bachelor of Technology (BTech), Computer Science Engineering (AI & ML)
VIT Bhopal University | Jul 2025 – Jul 2029

Senior Secondary — Physics, Chemistry, Maths & Computer Science
Aligarh Muslim University | Apr 2025 | Grade: 80% · Distinction in Four Subjects

An engineer dedicated to translating raw data into actionable intelligence. Specializing in agentic AI systems, LLM orchestration, quantitative finance, and scientific machine learning. Focused on building transparent, interpretable, production-grade systems that bridge complex algorithms and real-world impact.

Currently exploring:

• Advanced agentic architectures: LangGraph multi-agent consensus systems
• Retrieval-Augmented Generation and long-context LLM reasoning
• Physics-Informed Neural Networks and scientific ML
• Mechanistic interpretability and trustworthy AI
• Cloud-native ML infrastructure and scalable AI applications

English — Full Professional
Hindi — Native
Urdu — Native
German — Limited Working Proficiency (actively learning)

• GitHub — github.com/shamiquekhan
• LinkedIn — linkedin.com/in/shamique-khan
• Email — shamiquekhan18@gmail.com
• Quant ML — quantml.tech

Philosophy: Transparency in Technology
Approach: Data-Driven, Methodical, Iterative
Goal: Building intelligent systems that augment human capability