Portfolio

Nasar Siddiqui

SWE & Philosophy // USASK

Home

Experience

Projects

Skills

Contact

RESUME
Product Management Download PDF Software Engineering Download PDF

Available for freelance work

AVAILABLE NOW

Nasar

Siddiqui

SWE & Philosophy // USASK

I'm a Computer Science and Philosophy student at Usask. I build AI-assisted systems, full-stack products, data-integration pipelines, and custom software for civic, public-health, and business workflows. When I'm not coding, you can usually find me on a hike I was wildly overconfident about.

AI / Full-Stack / Systems

Experience

Founder & Full-Stack Developer

Apr. 2026 - Present

Civic Voice

  • Designed and built a cross-platform React Native/Expo product that turns fragmented Canadian legislative and policy data into personalized workflows for bill discovery, interest-based recommendations, representative feedback, and notifications.
  • Designed an evaluation system for AI-assisted legislative summaries, analyzing 1,500+ outputs across 500+ documents and 6 model/prompt configurations; converted 75+ failures into pre-launch regression tests, production quality monitoring, and runtime guardrails.
  • Engineered PostgreSQL/Supabase ingestion pipelines for heterogeneous government sources, with scheduled refreshes, normalization, validation, caching, freshness tracking, and fallbacks that protect users from stale or malformed data.

Featured

Projects

Encrypted Inference Gateway

A privacy-preserving ML inference gateway with formal API contracts, encrypted request handling, and reproducible benchmark tooling.

Python FastAPI Pyfhel OpenAPI 3.1 JSON Schema GitHub Actions Jupyter Notebook
More details
  • Designed and implemented a privacy-preserving inference gateway for ML inference with formal JSON Schema and OpenAPI 3.1 contracts to ensure wire-level correctness, interoperability, and safe client-server workflows.
  • Built layered validation pipelines for CKKS-based encrypted inference, separating schema, envelope, and ciphertext checks to enforce safe deserialization and parameter compatibility before execution.
  • Developed a Pyfhel-based reference backend and Python client SDK for model discovery, local session creation, client-side encryption/serialization, inference submission, encrypted result retrieval, and local decryption.
  • Verified protocol behavior with end-to-end integration, ciphertext round-trip, and plaintext/encrypted agreement tests, then built reproducible benchmark tooling and GitHub Actions CI to quantify latency, throughput, payload growth, and output agreement while improving regression detection and release reliability.
  • Optimized the encrypted inference path by caching reusable CKKS contexts across requests, reducing mean encrypted inference latency from ~311.9 ms to ~85.0 ms and improving end-to-end throughput by approximately 2.5×.

Civic Voice

A full-stack civic-information platform and mobile application that aggregates, validates, summarizes, and serves official legislative, policy, funding, and representative data across Canadian jurisdictions.

Expo React TypeScript Supabase PostgreSQL Edge Functions Cron OAuth
More details
  • Architected and built a full-stack civic-information platform aggregating official legislative, policy, funding, and representative data across multiple Canadian jurisdictions.
  • Designed a PostgreSQL/Supabase data layer with scheduled refresh jobs, serverless ingestion functions, public read views, caching, access controls, and source-freshness tracking.
  • Built resilient data-integration pipelines with normalization, validation, citation preservation, and safe fallbacks for incomplete or unreliable official sources.
  • Developed an AI-assisted summarization system using local Ollama models to convert official legislative text into accessible plain-language bill summaries.
  • Created an evaluation harness covering 500+ legislative documents and 8 quality criteria, then analyzed 1,500+ generated summaries using 12 structured validation checks.
  • Delivered a cross-platform Expo/React Native application with authentication, saved content, preferences, typed domain models, and server-driven push-notification workflows.

Sunny & Safe Saskatchewan

A public-health website that turns sun-safety education into clear resources, campaign storytelling, and practical sunscreen awareness tools.

TypeScript React Responsive UI Education
More details
  • Built a CMS-powered public-facing outreach site for Sunny & Safe Saskatchewan to make sun-safety information easier to find, understand, and act on.
  • Structured the site around campaign storytelling, educational resources, practical prevention guidance, and accessible contact pathways.
  • Added interactive, user-facing tools and content areas that help visitors connect sunscreen habits with real-world UV protection and public health awareness.

AI Engineering

RAG & Reranking, LLM Evaluation, Guardrails, Drift Monitoring, Prompt Testing

Languages

Python, SQL, Java, JavaScript, TypeScript

Backend & Systems

FastAPI, Flask, Node.js, REST APIs, MySQL, Supabase Edge Functions, OAuth

Data, Contracts & Infrastructure

PostgreSQL, Supabase, OpenAPI 3.1, JSON Schema, Cron, Docker, GitHub Actions, Monitoring

Philosophy & Writing

Ethics, Epistemology, Political Philosophy, Philosophy of Science, Technical Writing, Argumentation

Let's work together

Contact

Email nasar.siddi@gmail.com LinkedIn nasarsidd