Neelabho ChakrabortyAfter Hours

● After Hours · An interactive portfolio

NeelabhoChakraborty

  • AI / ML Engineer
  • Independent Researcher
  • Developer

I’m an AI and software engineer who builds products from the first sketch to the release, including native Mac apps and AI that runs on your own machine. I care about who ends up using them, and I like software that does its job and then leaves you alone.

View résumé Download PDF ↓

01 · Selected work

Things I’ve built

Pick a screen in the room, or a project here.

Open source

  • Fix additive seasonal initialization in Holt-Winters

    River (online machine learning) · Merged · November 2026

    A fix to how the Holt-Winters forecaster sets up its additive seasonal pattern, with regression tests and a before-and-after accuracy check. Merged by the maintainer.

Other work

  • AI Reel Studio Local AI studio

    A complete AI production studio running on a single Mac. I tuned and connected open-weight models for image, video, voice and lip sync into one automated workflow, from a script to stunning, finished vertical video. For ads, it turns a product page into scored hooks, scripts, storyboards and social copy.

  • SchoolDesk Desktop app

    Makes complicated Excel workbooks easy, for educators and for anyone who lives in spreadsheets. Edit marks and records, catch errors, build your own calculations, link sheets, then export a faithful Excel file and a clean PDF. It works the way you set it up, and everything stays on your computer.

  • Presentation Maker In use

    A desktop app for macOS and Windows that generates presentations with dynamic or rule-based layouts, whichever the person needs. Built for a working counsellor, who uses it.

02 · Research

Reading papers by rebuilding them

I replicate papers to understand them properly. The test is simple: can my own code get close to the number in the paper?

Replications will be listed here as they are finished, each with the paper’s number beside mine.

Interests

  • Applied machine learning
  • Local and on-device AI
  • Small and large language models
  • Model Context Protocol
  • Data science
  • Human-computer interaction
  • Privacy-conscious software

03 · About

Oman, then Canada, then India

I grew up in Oman, lived and studied in Canada, and came back to India. The people I met in each place wanted much the same from their tools: help with the job, without giving up their attention or their data.

In Canada I studied Computer Science and Mathematics at the University of Manitoba. I then chose to specialise, and now study Data Science and AI Development at BITS Pilani, alongside Computer Applications at Techno India University.

I like decisions backed by data, open-source models I can inspect and run myself, and software that respects people’s privacy.

View résumé Open PDF ↗ Download ↓

Skills

Programming

  • Python
  • Swift
  • TypeScript
  • JavaScript
  • SQL
  • Rust
  • Java
  • C++
  • C

Machine learning and data science

  • scikit-learn
  • pandas
  • Hugging Face
  • Open-source models
  • XGBoost
  • LightGBM
  • PyTorch
  • SHAP
  • Spatial cross-validation

LLM infrastructure

  • Ollama
  • MCP integration
  • Qwen 3 8B
  • MLX
  • Local LLM deployment
  • RAG
  • Agentic workflows
  • Tool calling
  • Model orchestration
  • Context engineering
  • Structured outputs
  • Structured extraction
  • LLM guardrails
  • Streaming inference
  • Inference optimisation

Generative and deep learning models

  • ComfyUI
  • FLUX.2
  • Wan2.2
  • Qwen3-TTS
  • MuseTalk

Data engineering

  • PDF parsing
  • OCR
  • Supabase
  • SQL
  • MinerU
  • GeoPandas
  • rasterio
  • xarray
  • DuckDB
  • Parquet
  • Report automation
  • PaddleOCR-VL 1.5
  • Surya OCR 2.5
  • Nanonets OCR2 3B
  • Granite-Docling
  • olmOCR 2
  • dots.ocr

Apps and interfaces

  • SwiftUI
  • AppKit
  • React
  • Tauri
  • Electron
  • FastAPI
  • MapLibre GL
  • ECharts
  • Tailwind CSS

Development tools

  • Git
  • GitHub Actions
  • GitHub Pages
  • FFmpeg
  • Twilio
  • pytest
  • uv
  • Ruff

■ What I build with ■ What I’ve shipped with ■ What I’m trying

04 · Experience

Work, study and community

Education

  • BS, Data Science and AI Development

    Birla Institute of Technology and Science, Pilani · Studying

  • Bachelor of Computer Applications

    Techno India University · Studying

  • Computer Science and Mathematics

    University of Manitoba, Canada

Community

  • Outreach volunteer University of Manitoba

    Led outreach that introduced school students to simulation and data.

All activity on GitHub

05 · Contact

Write to me

If you’re working on something like this, I’d like to hear about it.

No analytics and no cookies. The room, the people in it and the music are generated by your own device.