Skip to content
View DhruvB100's full-sized avatar

Achievements

Achievement: YOLOAchievement: Pull Shark

Achievements

Achievement: YOLOAchievement: Pull Shark

Highlights

  • Pro

Block or report DhruvB100

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
DhruvB100/README.md

Dhruv Bhatt

BSc (Honors) Computer Science @ University of Alberta · GPA 3.7 · Graduating 2027

I build things end-to-end: AI agents and LLM features, cloud infrastructure, data pipelines, backend APIs, and the React frontends on top. I care about systems that are correct, tested, and honest about what they don't know, and I measure them instead of guessing.

Open to: Winter 2027 co-ops and 2027 new-grad roles in software, data, AI/ML, and cloud. Edmonton, Calgary, remote in Canada, or open to relocating.

What I've Built

  • FilingsPilot: AI research agent over the SEC annual reports of 10 Canadian companies (RBC, TD, BMO, CIBC, Shopify, Suncor, and more). A LangGraph agent uses hybrid search (pgvector + keyword), guarded text-to-SQL, and a calculator tool, and cites every fact to a page or data row. If the filings don't contain the answer, it says so. Measured on a 50-question golden set: 96% correct, 100% faithful, 95.5% citation accuracy, about $0.003 per question. Guardrails against prompt injection and unsafe SQL, red-team tests in CI, and planned in Jira with decision records in Confluence. Solo project. I built the Python backend, agent, and React/TypeScript UI; Claude Code helped build the evals and generated most of the Azure/Terraform setup and the MCP server.

  • DataLens: Serverless data lakehouse on AWS. Ingests live USGS earthquake data every 6 hours (EventBridge + Lambda), lands it as partitioned Parquet in S3, catalogs it with Glue, and serves it two ways: a React/Recharts dashboard over Athena, or a plain-English question box where an LLM (Gemini) writes SQL that is safety-checked before it runs. Fully defined in Terraform, CI via GitHub Actions, CloudWatch alarms. Built with Claude Code set up as a reviewer, not an author; 18 real issues and fixes logged in the repo.

  • TestBench-CI: A Flask app plus its full regression suite: unit, integration, contract, SQL, and Selenium end-to-end tests using the Page Object Model, run in Docker Compose and in CI on every pull request with screenshots, logs, and JUnit XML on failure.

  • QueueWise: Full-stack waitlist and appointment platform. FastAPI + React/Vite/Tailwind, real-time queue dashboard over WebSockets, multi-tenant orgs, JWT auth, RBAC, auto-generated OpenAPI docs.

  • SQL-Server ETL Pipeline: Python ETL + T-SQL stored procedures + Power BI dashboard + automated PowerPoint reporting, with audit logging throughout. Replaced a manual monthly reporting process (~80% less effort).

  • F1 Race Strategy: ML pipeline over FastF1 telemetry. Regression + classification with GroupKFold CV (MAE ~0.32s, F1 ~0.76), Streamlit dashboard, unit tests, CI.

Currently

  • Looking for Winter 2027 co-ops and 2027 new-grad roles
  • Taking Intro to NLP and Cryptography for Digital Privacy
  • Sharpening algorithms and data structures (NeetCode roadmap)
  • 1st Place, University of Alberta Programming Contest (Division II), 2025

Stack

Python TypeScript JavaScript Java C Bash Script React FastAPI Flask LangGraph Pandas NumPy scikit-learn PyTorch Postgres SQL Server AWS Azure Terraform Docker GitHub Actions Git Jira Power Bi Streamlit

Contact

LinkedIn Email

Pinned Loading

  1. FilingsPilot FilingsPilot Public

    AI agent that answers questions about 10 Canadian companies' SEC annual filings with page-level citations. LangGraph + Azure OpenAI, hybrid RAG on pgvector, guarded text-to-SQL, 50-question eval su…

    Python

  2. DataLens DataLens Public

    Serverless data lakehouse on AWS that ingests earthquake data on a schedule, stores it as Parquet in S3, and lets you query it with SQL or plain English via an LLM. Built entirely with Terraform.

    HCL

  3. RL-algorithms RL-algorithms Public

    Q-Learning, SARSA, and DQN implemented from scratch and benchmarked across 4 OpenAI Gymnasium environments — tabular vs. neural network approximation, with hyperparameter tuning and reward shaping.

    Python

  4. QueueWise QueueWise Public

    Full-stack waitlist & appointment management system — real-time queue dashboard, multi-tenant orgs, JWT auth, WebSockets, Docker. Built with FastAPI + React.

    Python 1

  5. F1-Race-Strategy F1-Race-Strategy Public

    ML pipeline predicting F1 lap times & pit stop windows from FastF1 telemetry — XGBoost models, GroupKFold cross-validation, Streamlit dashboard.

    Python

  6. Vision-Based-Object-Detection Vision-Based-Object-Detection Public

    Classical (HOG+SVM) vs. deep learning (CNN) object detection pipeline with 2D pose estimation via heatmap regression — built from scratch in PyTorch and OpenCV.

    Python