Gatlat

I'm Gatlat

Computer Science & Psychology at McGill

First-generation student and developer. For two summers I've been working under Dr. Brian Chen at the McGill University Health Centre, on the web side of two of his lab's projects.

--:--
--
local time

Work

Enzygent

Infrastructure, Deployment & Interface Design · Research Institute of the McGill University Health Centre (RI-MUHC), Dr. Brian Chen Lab

May 2026 – Present

A retrobiosynthesis and enzyme design platform. Given a starting molecule and a target, it searches curated reaction databases for an enzyme already known to catalyze that transformation, and where none exists, generates candidate protein sequences for it.
Built the FIND engine: a chemistry matcher running Rhea and RetroRules in parallel with tiered precedence, behind a confidence gate that classifies every match as documented, lead, rule, or predict. An atom-balance and EC-consistency gate rejects matches whose enzyme class contradicts the net atomic change, and multi-step pathway search resolves single-intermediate routes with an identity guard. Gate thresholds tuned against a held-out benchmark harness rather than by hand.
Built the Design Studio for de novo design: magic-link and JWT authentication, Postgres-backed governance with sign-off from a second researcher, RFdiffusion and ProteinMPNN adapters behind feature flags on GPU endpoints, ESMFold fold-back for predicted structures, synchronized dual 3D viewers for candidate comparison, and a local-alignment homology screener gating every generated sequence.
Run the production server on lab hardware: three systemd units under an unprivileged account with the backend bound to loopback, fronted by a Cloudflare named tunnel that dials outward, since the hospital network blocks inbound connections and the address has to survive the machine relocating. Reboot-tested, with a runbook so the lab can operate it independently.
Debugged production at the layer the failure actually lived in — WAL journaling to end lock contention between a long index warmup read and concurrent writes, provisioning the EC/Rhea catalog on a box where the gitignored one-time import had never run, RDKit charge normalization after neutral user input silently failed to match charged physiological forms, and CORS headers on 500 responses so the browser surfaced real exceptions instead of "failed to fetch".
Designed the interface and froze it into design tokens and a shared component kit, then rebuilt the prediction screen, results page, history, and internal tools on it — answer-first results, a pathway diagram, collapsible evidence sections, and colour-coded tags that always carry a written label.
Set up GitHub Actions CI running frontend builds and backend lint and tests on every push.
UbuntusystemdCloudflare TunnelGitHub ActionsFastAPIPythonNext.jsReactTypeScriptRDKitSQLitePostgreSQLMol*RFdiffusionProteinMPNN

Web Developer Intern — Genedig

Research Institute of the McGill University Health Centre (RI-MUHC) · Dr. Brian Chen Lab

Jun 2025 – Sep 2025 · Jul 2026 – Mar 2027

Refactoring Genedig, the lab's published genome browser (BMC Bioinformatics, 2015) for co-navigating DNA, RNA, and protein sequences, under contract through March 2027.
Diagnosed and fixed critical bugs across Genedig's PHP/JavaScript codebase, improving platform stability and reducing downtime for the research team.
Built a local XAMPP/MySQL development environment that streamlined onboarding and let the team reproduce and troubleshoot issues faster.
Demoed the working platform to the research group, walking through fixes and gathering feedback to prioritize next steps.
PHPJavaScriptAngularMySQLXAMPP

Projects

Springboard.io

Co-creator · MDN Web Docs Hackathon

Mar 2026 – Apr 2026

Co-built an agentic employee-onboarding platform end-to-end in a single afternoon sprint — finalist, top 5 of 16 teams at the McGill Data Network Hackathon.
Shipped a working product under extreme time pressure using Next.js 16 and the Vercel AI SDK (Gemini), prioritizing a demoable end-to-end flow over feature breadth.
Designed agentic workflows that automated access provisioning and generated personalized intro emails for new hires.
Architected a RAG pipeline so new hires could get context-aware answers from internal docs instead of hunting through wikis.
Next.jsVercel AI SDKGeminiRAGTypeScript

DNA Sequence Binding Analysis

Bioinformatics · McGill University

2024 – 2025

Built a Python/Bash pipeline to process ChIP-seq data and call peaks, identifying transcription factor binding sites across the genome.
Identified transcription-factor binding motifs from ChIP-seq peaks and characterized protein–DNA interaction patterns.
Automated peak-calling and visualization end-to-end, cutting manual analysis steps out of the workflow.
PythonBashBioinformaticsChIP-seq

Medical Diagnostic Support System

ML Developer · McGill University

2024 – 2025

Built a Python diagnostic tool that compares patient symptoms against a case database to surface candidate diagnoses.
Implemented a K-Nearest Neighbours classifier for multi-class disease prediction, validating accuracy across held-out test cases.
Pythonscikit-learnKNNPandas

Skills

Languages

PythonJavaScriptTypeScriptPHPJavaHTML/CSSSQLBashC

Frameworks & Libraries

ReactNext.jsFastAPINode.jsAngularTailwindCSSRDKitPandasscikit-learn

Infrastructure & DevOps

MySQLSQLiteGitLinuxCloudflare TunnelXAMPPVercel

Tools

GitHubVS CodeVim

Education

Computer Science & Psychology

McGill University · Montréal, QC

2024 – 2028

WUSC-SRP Scholarship recipient (2024).
Coursework: Object-Oriented Programming, Calculus, Web Development, Software Design, Operating Systems & Computer Systems.
“And what is more enthralling to the human mind than this splendid, boundless, coloured mutability!—life in the making? How strange it is, then, that we should be contented to take such small parts of it as we can grasp, and to say, ‘This is the true explanation.’
David Grayson · Adventures in Contentment