AM.

Machine Learning Engineer · AI Researcher · Halifax, NS

Akash Maity

Computer Science graduate from Dalhousie University focused on machine learning, generative AI agents, and applied research. Currently building an AIS vessel-trajectory prediction pipeline with the Ocean Tracking Network, and leading applied ML projects at the Dalhousie Machine Learning Society.

View Projects Résumé ↓
Scroll

Get to know me

I turn messy data into models people can trust.

I'm a Computer Science graduate of Dalhousie University (2023–2026) with a focus on machine learning, deep learning, and generative AI. My work spans reinforcement learning, sequence models, retrieval-augmented generation, and building autonomous AI agents — from research pipelines to production-facing tools.

I'm currently a research collaborator with the Ocean Tracking Network at Dalhousie, building a full AIS data pipeline for vessel trajectory prediction under the supervision of Prof. Gabriel Spadon. Alongside research, I lead applied ML delivery as Technical Project Lead for the Dalhousie Machine Learning Society.

Looking for full-time Machine Learning Engineering or Data Science / ML roles with a January 2027 start date.

2026 B.CS, Dalhousie University
5+ ML/AI research & applied projects
2 Hackathon builds shipped
Photo of Akash Maity

Where I've worked

Experience

01

Mobile Application Developer

Collect (Startup Project) · Halifax, NS

  • Led a team building a React Native (Expo) platform for local creators, owning technical decisions across Firebase authentication, Cloudflare R2 media storage, and third-party API integrations.
  • Shipped a full-featured application including a video-based project feed, real-time event hub, and persistent light/dark UI.
Jan – May 2026
02

Teaching Assistant

Dalhousie University · Halifax, NS

  • Supported first- and second-year students in a required Computer Science course, mentoring on programming fundamentals and problem solving.
Jan – Apr 2024
03

Technical Project Lead

Dalhousie Machine Learning Society · Halifax, NS

  • Lead end-to-end delivery of applied ML projects for external clients and research partners, defining technical scope and coordinating cross-functional student teams from data acquisition through deployment.
  • Designed and delivered a hands-on workshop, "Build Your First AI Agent," using the smolagents framework, and organized the DMLS Summer Hackathon 2026 across five Atlantic Canada industry problem statements.
May 2026 – Present
04

Student Representative

Dalhousie Machine Learning Society · Halifax, NS

  • Serve as liaison between the student community and executive leadership; co-organize workshops, hackathons, and mentorship programs.
Nov 2025 – Present

What I've built

Projects

01
Nova Scotia Hospitals

ER Wait-Time Prediction Model

Built and deployed a machine learning model to predict emergency-room wait times, achieving an 18.5% reduction in wait times across Nova Scotia hospitals.

PythonMLHealthcare
02
50 arXiv RL Papers

RAG Hallucination Detection

Built a retrieval-augmented generation and faithfulness-evaluation system using FAISS vector search, Hugging Face sentence-transformers, and TinyLlama for generation, with an NLI-based detector that flagged unsupported claims.

RAGFAISSHugging Face
03
VRPTW

Multi-Agent Delivery Routing

Designed a multi-agent system with LLM-based intent extraction to convert natural-language delivery requests into structured routing constraints, reaching 100% constraint satisfaction after iterative prompt optimization.

LLM AgentsOptimizationPrompt Engineering
04
LendingClub Loan Default

Explainable ML Audit

Audited a LightGBM loan-default model using SHAP, LIME, and partial dependence plots, delivering case narratives and a business-facing slide deck that gave non-technical stakeholders a clear, actionable view of model risk.

SHAPLIMEExplainable AI
05
n8n AI Agent

Student Personal Assistant

Built a personal AI agent workflow in n8n that drafts, creates, and sends Gmail messages, tracks a budget, manages Google Calendar events, and answers general and health-related queries end-to-end.

n8nAutomationLLM
06
PSI & Watt the Hack

Hackathon Builds

PSI — a marine digital twin platform for offshore wind siting, built at the ShiftKey Labs Hackathon. Watt the Hack — a sustainability web app built with React, TypeScript, and Supabase.

ReactTypeScriptSupabase

That's the highlight reel.

More on GitHub →

Current focus

Where will the ship
be next?

Ocean Tracking Network (OTN), Dalhousie University · with Prof. Gabriel Spadon · Halifax, NS · 2026 – Present

  • Build and maintain a full AIS data pipeline for vessel trajectory prediction: parser, sliding-window trajectory batching, spatial filtering, an API ingestor, an evaluation-metrics module, a visual debugger, and a synthetic data generator.
  • Implement sequential motion-prediction stages — a Kalman filter baseline followed by a configurable GRU model with a time-decay gate for irregular AIS ping intervals — informed by a literature review of trajectory-forecasting architectures.
  • Collaborate with faculty supervisor Prof. Gabriel Spadon via Git-based version control and code review.
AIS DataKalman FilterGRUTime-Series

Illustrative sketch — predicted (pink) vs. actual (teal) motion. Not live model output.

Toolbox

Technical Skills

01

Languages & Frameworks

PythonSQLJavaScriptTypeScript
02

Machine Learning & Deep Learning

Supervised / Unsupervised Learning Reinforcement Learning Q-Learning / SARSA / DQN / CQL / PPO Actor-Critic Transfer Learning RNN / GRU / LSTM Kalman Filtering Recommender Systems ANN Semantic Search
03

Generative AI & Agents

Autonomous AI Agents LLM Fine-Tuning RAG (LangChain, FAISS) Hugging Face Transformers NLI Hallucination Detection Prompt Engineering SHAP / LIME Partial Dependence Plots
04

Data Engineering & Cloud

ETL Design Pipeline Orchestration Feature Engineering Synthetic Data Generation AWS Google Cloud Platform Firebase Cloudflare R2 Power BI REST APIs

Get in touch

Let's build something
worth shipping.

📄 Resume 📋 Full CV
Copied to clipboard