Open the playable cube + hoverable work tree →
Mithilesh Adhinarayanan
CSE · AI & DS SASTRA University Hosur, India
I like problems with a state, a move, and a measurable result.
The Rubik's cube is a useful model for that:
state ── move ──> new state
│ │
└── constraint ───┘
In code, that becomes reinforcement learning, data remediation, traffic control, forecasting, and the mathematics behind a decision.
π(a | s) choose an action from a state
P(s' | s, a) model the next state
Σ error(t) measure what needs fixing
| Direction | Current focus |
|---|---|
| Reinforcement learning | PPO, DQN, multi-agent control, sequential decisions |
| Quantitative finance | Time series, alpha research, risk, portfolio optimisation |
| Applied ML | Network anomaly detection, computer vision, data quality |
| Systems | Python APIs, dashboards, mobile apps, edge and data pipelines |
| Repository | Short description |
|---|---|
| Adaptive Traffic Management | A multi-agent traffic platform with computer-vision telemetry, DQN signal control, emergency preemption, routing, MQTT, and a safety layer that learned policies cannot override. Built from the SIH 2025 idea. |
| CraftHaat | An offline-first Flutter + FastAPI prototype for artisan listings from a voice note and a photo, with speech, image, catalog, and ONDC preparation steps. |
| Exchange Rate OpenEnv | An agent environment for deciding what to do with missing, spiking, or stale exchange-rate observations. |
| Deep Learning Projects | Foundational models and notebooks from deep-learning coursework. |
| EFSAT | Efficient Satellite Analyser project for SIH 2026. |
| PyMC / GSoC Preparation | Preparation and open-source contribution track for a PyMC-focused GSoC 2026 path. |
More original repositories: Amazon ML Challenge · Data Science Projects · BSTS · OSM Closure Routing PoC · ML4Sci prep · Prep · Portfolio source
The profile also contains forks used for learning and exploration; they are intentionally not labelled as projects created by me.
Working on a DNN + PPO approach to financial time series using macroeconomic signals such as crude oil, gold futures, and NIFTY 50.
DNN PPO time series quant finance
Investigating anomaly detection for Advanced Persistent Threat activity.
ML network security anomaly detection
Both are active research directions. The IEEE publication is a target, not a published result.
| Branch | Skills |
|---|---|
| AI / ML / deep learning | Reinforcement Learning · Deep Learning · PPO · DQN · MARL · Neural Networks · Policy Gradient · TensorFlow · PyTorch · Keras · Scikit-Learn · Backpropagation · Time Series · Transfer Learning · Model Optimisation |
| NLP / LLM | Natural Language Processing · Large Language Models · Transformers · BERT / GPT architectures · Hugging Face · Text Classification · Sentiment Analysis · Tokenisation · Embeddings · Prompt Engineering · RAG · LangChain |
| Computer vision | OpenCV · YOLO · CNNs · Object Detection · Image Classification · Image Segmentation · Real-time Video Processing · Feature Extraction · Vehicle Detection |
| Languages | Python · Java / Swing · C++ · C · SQL · JavaScript |
| Web / full stack | Django · Streamlit · Flask · FastAPI · React · HTML / CSS · MongoDB · REST APIs |
| Data / analytics | NumPy · Pandas · Matplotlib · Seaborn · Plotly · Feature Engineering · Data Cleaning · Statistical Analysis · EDA · Data Visualisation |
| Quantitative finance | WorldQuant BRAIN · Custom Expression Languages · Algorithmic Trading · Alpha Research · Financial Forecasting · Exchange-rate Modelling · Macroeconomic Analysis · Risk Analysis · Portfolio Optimisation · Backtesting · Probability |
| Tools / systems | High Performance Computing · Git / GitHub · Linux · Jupyter Notebook · Google Colab · VS Code · Docker · Raspberry Pi · DSA · APT Detection · Anomaly Detection · Network Security |
| Mathematics | Probability Theory · Linear Algebra · Calculus · Olympiad Mathematics · Physics · Statistical Modelling · Problem Solving |
- 2026–present — BRAIN Research Consultant, WorldQuant
- 2026–present — Freelance AI/ML developer
- 2026–present — Student researcher, SASTRA Deemed University
- 2025 — CMI STEMS Top 30 India
- 2025 — Smart India Hackathon project: intelligent traffic management
- IOQM — Qualifier and merit certificate holder
- Yale / Coursera — Financial Markets certification
The full tree has hover highlighting in the interactive lab. The README version is deliberately static because GitHub strips JavaScript from profile READMEs.
The graph is generated from GitHub's contribution calendar and refreshed daily by GitHub Actions. It is not a hand-drawn activity claim.
Research collaborations · internships · freelance AI/ML work