Brandon Tong Github LinkedIn
☆ Make a difference ⚛ Magnify relationships ☇ Manifest meaning
Toggle non-professional commentary ■ DO NOT ENABLE IF RECRUITER ■
⚛ Work Experience
☇ Current
Forward Deployed Engineer, AI Voice Agents
Cresta.ai — B2B Voice Agents, backed by a16z & Sequoia, $1.2B — San Francisco (Remote) — Jan 2026–Present
  • Deploy real-time voice agents for F500 clients across voice and chat contact center operations.
  • Scoped and implemented new agent products enabling the 2 largest deals in company history.
  • Architected platform-level changes in voice engineering, multilingual capabilities, and end-to-end AI agent evaluation.
  • FDE hire No. 6 (team now 50). Shaped process and practices. Awarded Propoel26 Conference: Best FDE Team.
  • Led product roadmap to bilingual and multi-lingual agents cross-functionally: design, GTM, platform, and customer.
hire no. 6 out of 50. the team was 6 people when i joined. it's now 50. i'm choosing to interpret this as causation.  also "forward deployed" is a military euphemism. i ship things. sometimes weekly. sometimes daily. occasionally i am the thing being deployed.
⚛ Consulting
AI Strategy & Implementation Consultant
Lemay.ai — Toronto — Feb 2024–Jan 2026
  • Engagement Manager: full consulting lifecycle — proposals, exec discovery, technical implementation, handover. Closed $20M+ in signed SOWs: NATO drone program, Canadian defense RFP, Caribbean AI transformation.
  • Client-Facing Research Engineer: production AI systems (RL, CV, LLM) across healthcare, finance, logistics, and defense.
  • Parallel genetic algorithm achieving 5× speedup, projected 6-figure annual cloud cost savings (Scale AI grant).
  • Fine-tuned CV model for magnetic resonance defect detection — manual review time −60%.
  • LLM pipeline extracting clinical note data — task time −95%.
"NATO drone program" is a real sentence i get to say now. i stopped telling people what i do at parties. also: $20M in SOWs sounds impressive until you realise consulting is mostly getting executives to agree on what they already decided to do.
Product / AI Consultant
Wand Inspiration LLC — Singapore / Remote — Dec 2023–Dec 2024
Tech strategy through research, feasibility studies, and user testing for a SaaS platform. AI coach-participant matching system using classical NLP for low-compute deployment.
"classical NLP for low-compute" is a polite way of saying "they had no GPU budget." still worked. regex and embeddings carry more weight than people admit.
⚄ Research
Machine Learning Researcher
UBC Computer Science — Vancouver — May–Dec 2023
  • LLM Inference (Advisor: Dr. Gujarati) — 10–15% speedup over PyTorch cuBLAS for LLAMA-2 grouped query attention; IO-aware 4D GEMM on A100 via OpenAI Triton.
  • Neural Architecture Search (Advisor: Dr. Wang) — novel approximate Top-K activations; runtime −50%, accuracy 97%; bandit-based max inner product search in PyTorch.
i spent 7 months making a GPU go 12% faster. OpenAI Triton is what happens when someone decides CUDA isn't painful enough and makes it programmable in Python. i would do it again. my friends who went to the beach would not understand.
⚖ Co-op
Software Engineer Co-op, Edge Compute
Splunk — San Francisco — May–Aug 2022
Launched edge computing product in Go/Java, integrating AWS S3 with Splunk Edge Search Services. Enhanced GitLab CI pipelines with DAST, reducing security risks across deployment pipeline.
first big tech internship. san francisco was exactly what everyone said it would be, and also nothing like what anyone said it would be. edge computing is just regular computing but you tell people it's at the edge and somehow it becomes a product.
☂ Intern
Software Engineer, Distributed Systems
ZALORA Group — Singapore — Jul–Dec 2020
Microservice API gateway handling 100k+ daily requests, improvements in Go. Integrated recommendation microservices via Redis, SQL, memcache, Apache Thrift.
100k requests/day sounds like a lot until you realise that's ~1 per second. still, when it breaks at 1am on a sale day, everyone in the building knows your name. also worked at a yoga studio during this stint. @yogamovement. i am a multifaceted individual.
♕ Education
⚛ Graduate
Johns Hopkins University
M.S. Applied Computational Mathematics — Remote / Baltimore — Deferred
Intended focus: real-time series machine learning. Fully deferred pending current engagement.
deferred because cresta happened. ask me in 2027 whether this was discipline or procrastination. i genuinely cannot tell yet.
♕ Undergraduate
University of British Columbia
B.Sc Computer Science — 4.0 GPA (core) — Vancouver — 2024
NLP, Randomized Algorithms, Advanced ML, Real-Time Distributed Systems, Bioinformatics. President: UBC Yoga & Boxing clubs. Electives: Theatre and Screen Acting, Directing.
theatre was the best decision i made at university. turns out learning to inhabit a character is useful when you're presenting to a board of directors at 26. also boxing is just fighting with extra steps, and yoga is just suffering with better lighting. i was president of both clubs. this explains a lot about me.
⚄ Diploma
Temasek Polytechnic
Financial Technology — Singapore — Top Student
Data Engineering, Financial Economics, Social Psychology.
top student in fintech at 17. this is the origin story of someone who had no idea what they wanted to do and just kept winning things hoping it would eventually become a personality. social psychology has been more useful in my career than 80% of the CS courses. i'm not sure what to do with that information.