A senior team of 10–20 engineers shipping AI into production

Who is behind this

Afraz Khan is the founder and lead AI engineer. He has delivered more than twenty contracts at a 5.0 average rating, building production AI for healthcare and financial-services clients including OphyCare, Ace Money Transfer and EdgeCIO.

His work covers the systems described across this site: a LangGraph orchestrator unifying four clinical sub-agents behind one conversational API, a KYC pipeline pairing OCR with facial recognition that removed €40,000 a year of manual review, a graph-plus-vector retrieval platform that lifted query accuracy 40%, and an AI medical scribe that auto-fills more than ten mandatory clinical forms per consultation.

He writes the engineering posts on this site. Where a post makes a claim about accuracy, latency or cost, it comes from work he shipped.

LinkedIn GitHub Upwork Full author profile

Leadership

EpochC is a team of 10–20 engineers. The people below own delivery — you work with them directly, not through an account manager.

• 

  Afraz Khan

  Founder & Lead AI Engineer

• 

  Salman

  Co-Founder & Cloud Engineer

• 

  Awais

  CTO & Full-Stack AI Engineer

How we work

We staff projects with senior engineers rather than layering juniors under a manager, and we take on a deliberately limited number at a time. That constrains how much we can accept, and it is the reason the delivery record above holds.

Every engagement starts with a scoping conversation where we try to establish whether the thing you want built is worth building. If an off-the-shelf product solves your problem, that is the advice you will get from us — it costs us a project and saves you a great deal more.

What we ship is production-shaped: typed API boundaries, observability wired in from the first commit, evaluation sets that catch regressions, CI/CD, and confidence signals so the system escalates what it should not decide alone. Prototypes that cannot survive real traffic are not interesting to us and should not be interesting to you.

What we work with

• Python
• LangGraph
• LangChain
• GPT-4o
• RAG
• pgvector
• Neo4j
• MongoDB
• Redis
• PaddleOCR
• YOLOv8
• Deepgram
• FastAPI
• gRPC
• PostgreSQL
• Docker
• AWS
• GitHub Actions
• Langfuse

We choose the model and stack around your constraints — data residency, latency budget, compliance posture, existing infrastructure — rather than fitting your problem to a default.

Work we have delivered

• Ace Money Transfer · FinTech

  KYC onboarding, fully automated

  €40k annual cost removed · 98% field-detection accuracy · 70% less compute vs. baseline

• OphyCare · Healthcare

  One conversational API over four agents

  4 sub-agents unified · 80% code cut in refactor · 0 citation hallucinations

• EdgeCIO · FinTech

  Graph + vector retrieval platform

  +40% query accuracy · +35% answer relevance · <200ms end-to-end latency

• OphyCare · Healthcare

  AI medical scribe & form autofill

  10+ forms auto-filled · real-time speech-to-text