λ
Lam Pham
Lam Pham
Small tools, large systems. AI that handles real money.
I build composable AI infrastructure for banking at Cake by VPBank — ML models serving 6.5M+ users, deep learning in production, voice agents handling 25M+ real calls. Before banking: shipped AI across telco, FMCG, mobility, gaming, and finance. Probably the only person in Vietnam who's built production AI across all of them.
Chief Data Science & AI Officer at Cake by VPBank, Vietnam's largest digital bank. I lead a 25-person DSAI team running ML models for 6.5M+ banking users, deep learning systems in production, and voice agents that have handled 25M+ real customer calls — credit scoring, revenue forecasting, causal inference, agentic workflows processing real money.
Before Cake: co-founded TrueID at VNG (Vietnam's first tech unicorn) — built a $1M+ ARR AI business serving 10M+ users (B2B2C). Before that: shipped production AI across every major sector — telco at Viettel, FMCG, mobility, gaming at VNG, financial sentiment at Sentifi. Studied computer science in Saint Petersburg for seven years. Kaggle Master. WSDM Cup winner. Tennis player, solo guitarist, father of two pianists (ABRSM Grade 5+).
I might be the only person in Vietnam who has built production AI across telco, FMCG, mobility, gaming, and finance — and also won international ML competitions, co-founded a product company, and now runs AI infrastructure at a bank. Each piece of that sentence is load-bearing.
Now
Building a multi-agent orchestrator (codename Rach) — five digital twin sub-agents as personal intelligence infrastructure. Shipping a voice agent evaluation playbook for our 150+ agent fleet with SNR monitoring, golden QA eval sets, response mode systems. Running an intel-brief pipeline that compresses Vietnamese financial reports from $8 to $0.80 per report via PDF→markdown→SQL cascade.
Teaching causal inference to the DSAI team using credit risk scenarios. Designing the Cake NextGen AI Banker program. Iterating on the IPO narrative where AI infrastructure is the core valuation driver.
Updated April 2026
Work
Tools
How I Build
Unix
Small composable tools that multiply force. One tool does one thing. Pipe them for complex behavior.
Slope > Intercept
Hire for learning rate, not current knowledge. Build for compounding. Applies to people, models, and systems.
Farmer
Optimize conditions, nurture without forcing, accept seasonal variability. Patient cultivation beats forced growth.
Hard to Vary
Good explanations have parts you can't swap without breaking the whole. If an element is decorative, remove it.
Ship > Paper
A model in production beats a model in a notebook. The march of nines starts after the demo works.
6.5M+ banking users served by ML models in production at Cake
25M+ real customer calls handled by voice agents in production
10M+ users on eKYC & identity products built at VNG/TrueID
40+ BFSI clients across banking, insurance, and fintech
$1M+ ARR — B2B2C AI products, co-founded and shipped
5 industry sectors — telco · FMCG · mobility · gaming · finance
Career
2024 — Present
Chief Data Science & AI Officer
Cake by VPBank — Vietnam's largest digital bank
Lead 25-person DSAI team. ML models serving 6.5M+ banking users, deep learning in production, voice agents handling 25M+ real calls. Building agentic banking infrastructure — causal inference for credit risk, multi-agent orchestration, voice AI fleet management. Target: 70% AI workforce by 2027.
2019 — 2023
Co-founder, TrueID · Head of AI Products
VNG Corporation — Vietnam's first tech unicorn
Co-founded TrueID — digital identity platform and eKYC for financial services. Built a $1M+ ARR AI business serving 10M+ users across B2B2C. Learned what works when real money and real identity are at stake.
2015 — 2018
Lead Data Scientist
Viettel · Sentifi AG · VNG Corp
Production AI across telco (Viettel), financial sentiment (Sentifi AG, Switzerland), gaming & social (VNG). Built data science pipelines for social listening, fraud detection, NLP for Vietnamese text. The years that proved AI generalizes across sectors if you understand the fundamentals.
2012 — 2015
Deputy Head of Mobile · Co-founder, Saigonapps
GNT Inc · FPT Software · Saigonapps
Led mobile development at GNT — built Mobion Music, #1 on Japan App Store. Then enterprise software at FPT, then startup co-founder. Built the instinct for shipping products, not just models.
Credentials
1ST PLACE
WSDM Analytics Cup 2019 — Fake news detection (ByteDance). Modified BERT, 700+ teams, ACM conference.
KEYNOTE
IFC World Bank FIDN 2020 — "Implementing Digital Identity Platforms for a Digital Economy." Banking infrastructure meets AI verification at scale.
KAGGLE
Kaggle Master. CIKM 2017 (4th). CIKM 2018 (8th). AICovid Vietnam 2021 (2nd).
SPEAKER
PyCon SG 2016 — music2vec. TopDev 2019 — Productionizing AI. VietnamWorks Tech Expo 2016.
OSS
bert4vn & albert4vn — first-wave Vietnamese language models. VietAI open lectures (2017).
Education

MSc Computer Science — Saint Petersburg State University, Russia (2001–2008). Seven years shaped by the mathematical tradition and formal systems thinking.

MSc Business Information Systems — Vietnamese-German University (2010–2012).

Open Source
1st place — WSDM 2019 fake news classification
BERT pre-trained for Vietnamese
albert4vn
Lightweight ALBERT for Vietnamese (2019)
2nd place — COVID cough classification (2021)
Contact
Email
laampt@gmail.com
LinkedIn
lam-pham
GitHub
lampts
Kaggle
saigonapps
Location
Ho Chi Minh City, Vietnam