Curious engineer,
following the interesting problems.
Three years as an MLE, now an MS student at USC. The work has ranged from warehouse computer vision to LLM pipelines — different problems, same approach: figure it out while you're building it.
Cooking, games, films, sports, hiking, occasionally an instrument. Always something different, always all in.
Work
Tiger Analytics
Machine Learning Engineer
Aug 2021 – Jul 2024 · Remote
Profit Pool — Global Profitability Analytics
Python · PySpark · Databricks · Airflow · GCP · Power BI
- Cut a ~3-day manual reporting cycle to ~2 hours by rebuilding it as a fully orchestrated Airflow pipeline with automated data-quality gates.
- Built unified profitability dashboards across EU markets processing 26.7M retailer transactions (~200GB raw data), consumed by senior business stakeholders.
Price Elasticity Model — SKU-Level Pricing Intelligence
Python · Airflow · Dataproc · Vertex AI · BigQuery · Cloud Build · Docker
- Productionized a price elasticity pipeline across ~13,000 SKUs on GCP — 4-week cadence, outputs classified as inelastic / elastic / anomalous per SKU.
- Automated executive summary reports (top candidates, flagged SKUs) emailed to stakeholders each run; CI/CD via Cloud Build.
Perfect Pallet — Computer Vision for Warehouse Automation
Python · YOLOv5/v6 · PyTorch · OpenCV · Edge Deployment
- Deployed a YOLOv5 pallet inspection model on an edge device mounted on a live warehouse forklift — no GPU in production, mAP@0.5 ≈ 0.76.
- Replaced a manual pick-and-pack process, enabling live SKU counting, automated billing, and real-time order status updates.
Enverus
Business Technology Analyst, Intern
Feb 2021 – Jul 2021 · Remote
- Built a PDF extraction pipeline for oil and gas field receipts — OCR and layout detection on scanned documents, converting unstructured field paperwork into structured database records.
TagIT
Co-founder & CTO
2018 – 2021
- Built a modular theft-prevention chip that attaches to existing retail tags — real-time monitoring, alert system, and triangulation to pinpoint theft location in-store.
- Received a granted patent for the hardware design.
University of Southern California
Course Grader — Foundations of Database Management
Jan 2025 – May 2025
- Graded assignments and resolved queries for 200+ students across OS fundamentals, databases, Spark, RDD, NoSQL, and LLM-as-RAG.
Student Assistant, IMSC Summer Program
Jul 2025 – Aug 2025
- Designed and ran a hackathon for 80+ participants; built automated evaluation pipeline processing 3,000+ predictions per run.
USC — MS Applied Data Science
Aug 2024 – May 2026 (Expected)
GPA: 3.9 / 4.0
VIT Vellore — B.Tech ECE
Jul 2017 – Jun 2021
GPA: 8.85 / 10
"Identifying Stuttering Using Deep Learning"
IJITEE, Vol. 8, Issue 11, Sep 2019
"IoT-enabled chip to tag objects for theft prevention in retail stores"
Patent No. 2020041022886
Projects
LLM AutoFiller
Chrome extension that auto-fills job applications in under 3 seconds using your own API key. No backend, no server storage — everything stays in the browser.
Live Cooking Guide
Paste a TikTok, YouTube, or Instagram recipe URL and get a step-by-step guide with a live AI assistant. Ingestion under 30 seconds — scraping, transcription, parsing, all async.
Multi-Perspective Transformers for ARC-AGI-2
Built a ~20M parameter decoder-only transformer for abstract visual reasoning on ARC-AGI-2. Used multi-view augmentation and Product-of-Experts scoring to push consistency across puzzle orientations. 2nd place.
Research Assistant — Computer Vision
Model ensembling approach for object detection and tracking under occlusion, bad lighting, and adverse weather — fusing IR and RGB camera inputs. Building evaluation harnesses for cross-modality and cross-architecture comparison.
Yelp Recommendation System
XGBoost recommender with 43 engineered features — RMSE 0.9737 on a hidden test set. 142K predictions generated in under 4 minutes under strict Spark RDD-only constraints.
Contact
Let's build something.
Open to AI Engineer, MLE, and Data Science roles. If you're working on something interesting, I'm probably already curious about it.
veda.tibrewal@gmail.com