Sahil Rane's Portfolio

Experience

Writer

Medium · Self-employed

Nov '22 – Present

⤷ Writing blogs on philosophy, psychology, history & life experiences:) | Sometimes I write 𝒑𝒐𝒆𝒎𝒔.

Blog

Poems

SDE Intern

Godrej Infotech Ltd · Remote

Jun '26 – Aug '26

⤷ Developed a full-stack Non-Destructive Examination (NDE) Quality Testing platform using React, Node.js, and Express, managing industrial inspection workflows across RT, UT, MPT, and PT methods

⤷ Implemented dynamic data visualization dashboards with Recharts, full project CRUD management, and a multi-step inspection report wizard with draft auto-save and state restoration.

⤷ Designed and optimized NoSQL schemas in Amazon DynamoDB using Global Secondary Indexes (GSIs) for O(1) query lookups, preparing a serverless cloud architecture on AWS using AWS Amplify, AWS Lambda, and Amazon API Gateway.

Research Analyst Fellow

Deepcytes Cyberlabs (UK) · Remote

Jul '25 – Dec '25

⤷ Spearheaded frontend development for the "CyberSafe" CMS Platform for cybercrime intelligence and prevention.

⤷ Contributed to the Django backend architecture to support robust content delivery.

⤷ Optimized AI capabilities by refining the knowledge tree and engineering prompts for accurate data retrieval.

Proof-of-Work

QuantDevs – Options Pricing Engine

IIT G Winner. Institutional-grade American options engine (Binomial Trees + Numba JIT, <100ms). Features EWMA volatility forecasting and a LLaMA-3.3 AI Consultant that translates complex Greeks into actionable trade memos.

Python Numba JIT LLaMA-3.3 yfinance

S-TARE – Aviation Intel

Palantir-inspired situational awareness platform. Orchestrates live OpenSky telemetry, localized weather fusion, and NTSB-RAG (FAISS) via a tiered AI pipeline. Rendered in a high-fidelity CesiumJS 3D tactical view.

FastAPI CesiumJS RAG llamacpp

MarkovGPU – OpenCL Engine

High-performance library for massive-scale HMMs. Implements custom OpenCL kernels for Log-Sum-Exp stability, achieving 25x speedups over NumPy. CUDA-agnostic, supporting AMD, Intel, and Apple Silicon.

C++ / OpenCL Python HMM GPGPU

Advaita-Chita – State Steering

Research on Metacognitive Feedback Loops. Uses a Latent Centroid Observer to monitor hidden states of Qwen-2.5 in real-time, self-governing model entropy via a learnable Markov transition matrix.

PyTorch Qwen-2.5 Markov Chains Information Theory

Wolftor – STEM Visualizer

AI educator turning equations into Manim animations (3Blue1Brown style). Leverages Llama 3.3 for reasoning and Gemini 2.5 for visual scripting. FFmpeg-optimized backend for real-time synthesis.

Django Manim Gemini 2.5 FFmpeg

Clairis – Agentic Form AI

Serverless AI agent transforming rigid forms into natural conversations. Features Gemini Vision OCR, multilingual support, and an n8n-orchestrated decision engine. 100% local, privacy-first.

Gemini API n8n OCR Extension

Stack & Tools

Bachelor of Technology ~ IT

University of Mumbai · CGPA: 8.6

2023 – 2027

• Core Coursework: Maths, Data Structures & Algorithms, Database Management Systems, Computer Networking, Machine Learning, DevOps.
• Interdisciplinary Electives: Social Psychology, Fundamentals of Bioinformatics.

Pre-University

Ramnarain Ruia Autonomous College, Mumbai

2021 – 2023

HSC · Vocational Electronics · A grade

• Electronics: Digital & analog electronics, circuit design, PCB layouts, integrated circuits (ICs), Boolean logic, signal processing, op-amps, encoders/decoders, etc.
• Core Sciences (PCM): Quantum physics, classical mechanics, advanced calculus, inorganic & quantum chemistry, etc.

Project Brief

Foundations of Deep Learning

NPTEL · IISc Bengaluru

Pursuing

Core Architectures: Multilayer Perceptrons (MLP), CNNs (AlexNet, VGGNet, ResNet), Autoencoders, VAEs, and Recurrent Neural Networks (RNN, LSTM, GRU).
Advanced Models & Optimization: Transformers, LLM Pre-training & Fine-tuning, Diffusion Models, GANs, Backpropagation variants, and Explainable AI (SHAP, LIME, Grad-CAM).

CS50 Program

Harvard University · CS50

2025 – Present

Introduction to Databases with SQL : CRUD operations, joins, normalization, indexing, transactions & query optimization

Foundations of Machine Learning

Vizuara

Completed

Math for ML: Linear Algebra (Eigenvalues, Matrices), Probability & Statistics (Bayes, Distributions), and Multivariate Calculus.
Optimization & Implementation: Gradient Descent (SGD, Adam, RMSprop), Regularization, and programming algorithms from scratch alongside PyTorch & Scikit-Learn.

Sahil Rane साहिल राणे