Distributed Systems & AI

Vansh Taneja

Computing Science (AI) at the University of Alberta. I build event-driven platforms, backend infrastructure, and applied ML systems that scale.

Distributed Systems ML Systems Data Engineering LLM Pipelines

Building

Available for Winter 2027 Internships Edmonton, AB
Focus Software Engineering · AI/ML
Current ION — Certification Platform
Go C# Python AWS FastAPI Kafka

Case Study · City of Edmonton

ION — Instructor Operations Network

A production certification and competency-tracking platform for the City of Edmonton’s Waste Services Training branch. ION replaced spreadsheet-based tracking of operator certifications, annual competency assessments, and level-up paths across the city’s work areas.

Waste Services runs operators through 3-year certification cycles, annual competency assessments, and level-up training across a hierarchy of work areas. Tracking who is certified, on what equipment, and when it expires was manual and error-prone — and getting it wrong can put an uncertified operator on city equipment. ION models the whole lifecycle, including incident-driven recertification. Real app, real city operators.

Access control

Role-aware supervisor workflow

Supervisors see only their own work areas, with authority cascading down a parent→child work-area tree. A data-driven permission check gates the screen; a two-pane work-area layout drives operator and path loading on selection.

Assessment engine

Assessment builder integration

Wired a custom JSON-driven assessment engine into a full catalogue UI with active/inactive states, live save and validation, and a scoring model. Root-caused two subtle bugs: the engine only read its config on mount, and score calculation broke on serialization.

Path lifecycle

Automated path creation

Designed the rule that completing a certification course auto-creates the learner’s competency path (defaulted inactive), removing manual supervisor overhead — with duplicate-path guards per work area and multiple paths permitted across areas.

CSS architecture

Reusable stylesheet layer

Owned the global stylesheet: reusable classes in a clean split from screen-specific overrides, keeping the City of Edmonton design system consistent across every screen.

What I learned

  • Idempotent writes. Always reassign the returned Id after a CreateOrUpdate — skip it and repeated saves silently insert duplicates. I root-caused a double-create bug exactly this way.
  • Scope correctness. Reading .Current outside a list-row scope silently resolves to row 0. Pass the full record as an explicit parameter instead.

Certifications

Validated cloud expertise and credentials.

About

Engineering reliable software with AI in mind.

  • Build software that is reliable, scalable, and easy to maintain
  • Bridge backend engineering with practical AI/ML in production
  • Ship solutions that improve user workflows and system quality
  • Backend services, APIs, and event-driven pipelines
  • ML systems including NLP, RL, and LLM guardrails
  • Cloud deployments with Docker and AWS tooling
  • Co-op software developer at the City of Edmonton, shipping ION for Waste Services Training
  • Building distributed systems and AI projects outside of work
  • Open to Winter 2027 software engineering and ML internships

Skills

Tools, languages, and platforms I ship with.

LLM Guardrails Event-Driven Systems FastAPI Kafka RabbitMQ PyTorch Redis PostgreSQL OpenAI API

Languages

JavaC#GoPythonC++SQL TypeScriptJavaScript

Backend & Systems

Spring Boot.NETFastAPIFlaskREST APIsgRPCOutSystems 11

Cloud & DevOps

AWS (CCP)DockerGitHub ActionsJenkinsKafkaRabbitMQNginx

Databases

PostgreSQLMongoDBRedisMinIO

AI / ML

PyTorchTensorFlowScikit-learnXGBoostOpenCVOpenAI API

Concepts

Distributed SystemsMicroservicesCI/CDUnit TestingAgile/Scrum

Projects

Systems I’ve architected and shipped.

LiteLLM Python · Open Source

LiteLLM — Production Bug Fixes (Open Source)

Contributing fixes to LiteLLM, the widely used open source LLM proxy — currently in maintainer review.

  • Fixed a Gemini thought-signature bug causing HTTP 500s for Codex-style clients — independently verified by an affected user who retired their proxy-middleware workaround (PR #32545)
  • Expanded sentinel and access-group handling in the health endpoint (PR #31505)
  • Each fix shipped with regression tests and live-proxy proof
SocialDistribution Django · DRF · PostgreSQL

SocialDistribution — Federated Social Network Node

A node in a live federated network: four independent teams built servers from a shared API spec, and posts, likes, comments, and follows cross node boundaries. Contributor on an 8-person team.

  • Federates live with 3 independently built nodes — 663 automated tests, 29 REST endpoints
  • Owned author identity (FQIDs), entry distribution to remote inboxes, and the public API docs other teams integrated against
Software Quality Engineer Winter 2026 · CMPUT 402

Software Quality Engineer — Tartan Smart Home Platform

Designed test cases from system requirements and edge cases.

  • Validated REST services, simulator behavior, and state consistency
  • Documented defects and regression findings for fixes

Experience

Where I’ve built, taught, and shipped.

Co-op Software Developer — Rapid Development Services

Current · through Dec 2026 · City of Edmonton
  • Own the full front-end of ION, a production certification and competency-tracking platform for Waste Services Training
  • Built role-aware supervisor workflows, integrated a JSON-driven assessment engine, and automated the competency-path lifecycle
  • Maintain the reusable global CSS layer that keeps the City of Edmonton design system consistent across the app

Teaching Assistant — CMPUT 261 (Intro to AI)

Winter 2026 · University of Alberta
  • Led weekly labs for 200+ students covering search and reinforcement learning
  • Reviewed Python assignments for correctness and efficiency
  • Explained complex AI concepts with clear, structured feedback

Research Assistant — Data Engineering

Feb 2025 – Mar 2025 · University of Alberta (ACLMR)
  • Engineered fault-tolerant Python ETL pipelines processing 50M+ records
  • Partitioned datasets to meet API rate limits and memory constraints
  • Translated complex schemas into validated SQL datasets

Software Development Intern

May 2024 – Aug 2024 · Vita Actives
  • Deployed Flask microservices on AWS EC2 with Docker and Nginx
  • Built REST APIs with JWT auth and PostgreSQL dashboards
  • Implemented CI/CD with GitHub Actions and monitoring

Data Science Intern

May 2023 – Aug 2023 · Vita Activa
  • Optimized inventory forecasting using Random Forest and XGBoost
  • Built segmentation models with K-means and DBSCAN
  • Applied NLP to customer feedback for product insights

Education

University of Alberta — BSc Computing Science (AI).

Sep 2022 – Dec 2026

Edmonton, AB

Relevant Coursework

Data Structures & AlgorithmsSoftware EngineeringOperating Systems Database SystemsComputer VisionReinforcement Learning

Contact

Let’s build something ambitious.