"The ones who go the farthest are the ones who never stop building."

"What you can’t measure, you can’t improve. What you can imagine, you can build."

“Learn the rules like a pro so you can break them like an artist.” - Picasso

“Greatness is engineered, not inherited.”

"People who are crazy enough to think they can change the world are the ones who do."

I am an entrepreneurial-minded problem-solver, systems-oriented engineer and a technology-finance enthusiast, currently pursuing a Bachelor of Science in Mathematics and Computer Science at École Polytechnique (France).

My interests lie at the intersection of Machine Learning, Reinforcement Learning, Quantitative Finance, and high-performance software systems. I am particularly drawn to problems that sit at the boundary of theory and deployment, where mathematical modeling, algorithmic design, and real-world engineering constraints coexist. I am also almost equally buzzed about the intricacies of LLMs, Generative AI, Computer Vision, Financial Engineering and Hardware-Software Optimization.

I enjoy building end-to-end systems - from research prototypes and optimization pipelines to production-grade APIs, open-source libraries, and deployed platforms. I care deeply about correctness, scalability, interpretability, and measurable performance, and I prefer work that stands up to scrutiny rather than superficial demos.

Long-term, my goal is to bridge AI/ML, Mathematics, and Finance through a combination of research-driven exploration and product-focused execution, translating abstract ideas into systems that operate under realistic constraints.

• Bachelor Thesis (2026) - Reinforcement Learning for Maritime Navigation
  Graph-structured RL, GNN-based traffic representations, and safety–efficiency trade-offs.

• Building Open-source Tools and Platforms - See JBAC Edtech and keep out an eye for some upcoming repositories!

• Quantitative Strategy & RL Systems
  Portfolio optimization, cost-aware RL agents, and systematic evaluation frameworks.

• Open-source & Product Systems
  Designing, shipping, and maintaining production-ready ML and security tooling.

• Multi-objective Optimization Research
  Attempt at Analysis of evolutionary optimizers (COMO-CMA-ES) with benchmarking against baseline solvers.

Python • Static & Dynamic Analysis • RAG-assisted Security

• Open-source Python toolkit for detecting secrets, vulnerabilities, and misconfigurations across Git repositories, web applications, browser extensions, and network traffic
• Modular multi-scanner architecture including:
  • Git history scanning for committed secrets
  • Web crawler for endpoint and source discovery
  • Browser runtime inspection via Playwright
  • Network traffic inspection through a MITM proxy
  • RAG-assisted semantic analysis with CWE enrichment
• Includes 50+ built-in detection patterns for cloud providers and common services, with support for custom extensions
• Distributed as both a CLI tool and Python SDK, designed for flexible integration
• Achieved ~4,000 downloads, demonstrating real-world adoption and developer trust

FastAPI • Angular • AWS (Lambda, DynamoDB, EC2, Bedrock) • Finance/Trading

• End-to-end deployed platform for trading learning, feedback, and paper-trading simulation
• Real-time market-data engine computing indicators such as RSI, MACD, Bollinger Bands, SMA/EMA, and P&L
• Multi-agent (integrated) LLM architecture (Coach / Critic / Planner) powered by Amazon Bedrock
• Optimized for low latency, fast cold starts, and cost-efficient inference

CLI • Audio Processing • LLM Pipelines • Optimization

• Python library and CLI that converts educational YouTube videos into printable questionnaires
• Provider-agnostic pipeline supporting multiple LLM and transcription backends
• End-to-end workflow: audio extraction → transcription → summarization → question generation → PDF rendering
• Modern PEP-621 packaging, automated testing, and CI/CD via GitHub Actions
• 250+ downloads within the first 72 hours of release

Python • Stockfish • LLMs • LangGraph • Explainable Chess AI

• Explainable AI-based chess analysis system combining classical engines with LLM reasoning
• Integrates Stockfish for high-accuracy evaluation with natural-language explanations of moves and positions
• Modular pipeline for best-move recommendations, mistake detection, and strategic insights
• Uses LangGraph for structured multi-step reasoning and agent orchestration
• Distributed as a Python SDK and CLI tool for interactive and batch analysis
• Designed for training, education, and research with emphasis on interpretability over black-box evaluation

• End-to-end platform converting educational videos into structured summaries and printable questionnaires
• Automated pipeline: transcription → summarisation → question generation → PDF delivery
• Optimised GPU–CPU workflow reducing processing time by ~92% on long-form videos
• Supports large file uploads (400MB+) with secure authentication via Google OAuth
• Cloud-native architecture using GCP VM, AWS S3/DynamoDB, and Nginx reverse proxy
• Designed as a learning tool, emphasising speed, reliability, and accessibility

Python • FastAPI • Streamlit • LLMs • Voice AI • AWS • PostgreSQL

• Full-stack AI platform simulating real-time voice interviews with live transcription and automated feedback
• End-to-end pipeline: speech-to-text → question generation → response analysis → PDF report delivery
• Integrated Mistral Voxtral for transcription, GPT OSS-20B for evaluation, and gTTS for text-to-speech
• Secure authentication and user data management using bcrypt and PostgreSQL
• Deployed on Render with AWS S3 storage and CI/CD automation
• Optimised for free-tier infrastructure (512MB RAM, fractional CPU) to maximise accessibility

Python • Stable-Baselines3 • Gymnasium

• Designed cost-aware PPO agents for multi-asset portfolio rebalancing
• Explicit modeling of transaction costs, turnover control, and position constraints
• Achieved Sharpe ≈ 0.33 against multiple baselines
• Evaluation using bootstrap confidence intervals and Diebold–Mariano tests

C++ • Bitboards • OOP • React • Flask

• Modular chess engine built using bitboards and clean object-oriented design
• Implemented move generation, leaper logic, promotions, and debugging utilities
• Integrated with a React–Flask web interface for online play
• Ranked Top 10 / 25 in a competitive academic setting

Python • Backtesting • Optimization • Machine Learning

• Modular research and experimentation framework for quantitative strategy design and evaluation
• Emphasis on systematic testing under realistic constraints rather than over-fitted backtests
• Supports factor-based signals, portfolio construction logic, and performance analysis
• Serves as the experimental backbone for trading and RL-based projects within JBAC

Reinforcement Learning • Search • Heuristics

• Experimental platform exploring AI-driven problem-solving in structured game environments
• Focus on reward design, policy learning, and efficient state representations
• Used as a sandbox for reinforcement-learning ideas transferable to real-world optimization tasks

These projects shaped my foundations in algorithmic thinking and system design:

• Roommate Allocation Algorithm for incoming Bachelor students at École Polytechnique (constraint satisfaction; 150+ students, ~90% satisfaction)
• Mentor–Mentee Allocation Algorithm
• Hackathon Scraper (v1) | Repo
• Early ML pipelines (classification, PCA, ensemble methods) - including the course project for the Machine Learning course at École Polytechnique

Python · C++ · C · Java · Haskell · JavaScript · HTML/CSS · Bash · SQL(Foundational) · Prolog(Foundational)

PyTorch · TensorFlow · scikit-learn · OpenCV · NumPy · Pandas · Stable-Baselines3 · LLM APIs

FastAPI · Flask · React · Node.js · Express · PostgreSQL · MySQL · BeautifulSoup · REST APIs

Docker · AWS · GCP · Linux · Git · Nginx · CI/CD

Streamlit · Matplotlib · Tkinter

Developer Tooling: Jupyter · VSCode · IntelliJ · $\LaTeX$ · Markdown · GitHub · GitLab · Anaconda

Quantitative & Financial Systems -
Alpha strategy research · Portfolio construction and analysis · Monte Carlo simulation · Backtesting & frameworks · Black–Scholes modeling (Currently Learning)

AI/Machine Learning & Algorithms -
Supervised, Unsupervised & Reinforcement Learning · Neural networks · LLM-driven systems · Predictive modeling · Model evaluation · Feature engineering · Algorithm Design & Optimization (Currently Learning in Some Areas)

Mathematical Optimization & Applied Mathematics -
Linear & dynamic programming · Convex optimization · Numerical methods · Probability & (Asymptotic) Statistics

Backend Development & Systems (Engineering) -
Backend & API design/development · Distributed services · Cloud deployment · Authentication & security · Performance optimisation & workflows · CLI tooling (Bash/Zsh)

Data Engineering & Analysis
Data preprocessing · Data pipelines · Algorithms & data structures · Experimental design & evaluation

Experimental & Systems Engineering
Benchmarking · Reproducibility · Performance profiling · Resource optimisation · Secure system design

• 🌐 Websites: https://jaiansh.me | https://jbac.dev
• 💼 LinkedIn: https://linkedin.com/in/jai-ansh-bindra
• 📦 PyPI: https://pypi.org/user/Jbac_dev/
• ✉️ Email: jai-ansh.bindra@polytechnique.edu

Precision in thought. Boldness in execution.