Digaant DograGet in touch

ML/AI Engineer

Digaant Dogra

I build multimodal LLM systems and measure whether they work, from self-hosted inference on GPUs to the pipelines and tests around them.

Digaant Dogra in a cap and glasses, standing in front of a stone wall

Available full-time from January 2027

See work

About

I'm finishing an Honours Bachelor of Computer Science at Sheridan College. As an AI research assistant at Stereo Logic, I built vision-language model pipelines, moved inference to self-hosted models on an NVIDIA DGX, and built the evaluation tooling to benchmark them. I'm looking for an entry-level ML/AI Engineer role or internship.

Selected work

Case studies from applied research, my capstone and personal projects: the context, what I built and the tools I used.

More on GitHub

  • Vision-language models for spreadsheet activity

    Oct 2025 – Nov 2026

    Applied research on recognising business actions from screenshots and keyboard and mouse activity, using vision-language models and LLMs on self-hosted infrastructure.

    Stereo Logic, Sheridan

  • FootyLevels

    2026

    An iOS app that turns a player's phone video into a 3D motion model and overlays two attempts in 3D, so players can see where their technique differs.

    Capstone project

    300+automated tests

  • Chat with your documents (RAG)

    2025

    A retrieval-augmented generation system that answers questions from a document collection with local models, grounding answers in the documents and removing LLM API costs.

    Personal project

    +45%answer accuracy vs. a standalone LLM

Experience

  • Work

    AI Research Assistant / ML Developer

    Stereo Logic, Sheridan · Oakville, ON
    –

    • Worked on applied research into recognising business actions, such as text entry, field selection and navigation, from screenshots and mouse and keyboard event streams, using vision-language models and LLMs.
    • Extended the event pipeline to spreadsheets, adding prompts for clicks, typing, drag-and-drop and paste.
    • Migrated inference from OpenAI's API to self-hosted multimodal LLMs (Gemma 4 31B, Qwen3.8-27B, Qwen3-VL, Qwen3-Omni) on vLLM and an NVIDIA DGX, and built tooling to evaluate them.

  • Volunteer

    Technological Team Member (Volunteer)

    Google Developer Groups Sheridan · Oakville, ON
    –

    • Integrated and maintained 10+ internal tools and platforms, improving operational efficiency by 25% and reducing downtime during events.
    • Implemented role-based access control (RBAC) across club systems, ensuring 100% compliance with access protocols during member onboarding and offboarding.
    • Supported 5+ major tech events and workshops with cross-functional teams, helping boost member participation by 30%.

Education

  • –

    Honours Bachelor of Computer Science (Mobile Computing)

    Sheridan College · Oakville, ON

Skills

Generative AI & LLMs

  • Claude (Opus, Sonnet)
  • GPT-4
  • Gemini
  • Gemma
  • Qwen 3 series
  • Model Context Protocol (MCP)
  • LangChain
  • LangGraph
  • RAG
  • vLLM
  • CUDA
  • Fine-tuning concepts

Software engineering

  • Python
  • FastAPI
  • Flask
  • Java
  • C#
  • JavaScript
  • REST APIs
  • System design

Cloud & MLOps

  • AWS
  • Google Cloud
  • Docker
  • Docker Hub
  • Kubernetes
  • Terraform

Data

  • SQL Server
  • MySQL
  • MongoDB
  • Supabase
  • Firebase

Developer tools

  • Git and GitHub
  • Hugging Face
  • Jira
  • Confluence

Contact

Available full-time from January 2027. Get in touch by email or on LinkedIn.