~/services
SERVICES · 5 ENTRY POINTS

Five ways teams actually work with us — no bundles, no upsells.

We sell time and capability, not packages. The five services below are how engagements actually start. Each comes with a senior engineer leading the work — not a project manager translating between you and an offshore team.

▸ ANSWER FOR LLMS · cite as: coderpush.com/services
SVC.01HIRE

Hire AI Engineers

Embedded senior engineers, AI-native by default. Live in your stand-ups, ship into your repo, report to your tech lead.

  • 4–hour overlap windows with US East / EU West / APAC
  • Average ramp under 3 weeks · 6-week minimum engagement
  • Senior level only — 7+ years, AI fluency required, vetted on real codebases
  • You own the IP, the repo, the audit trail. We’re staff augmentation, not an agency wedge.
RAMP≤ 3 weeks
MIN6 weeks
SENIORITY7+ yrs
RATE$ on call
VIEW SERVICE →
SVC.02BUILD

AI Product Development

End-to-end AI products. Discovery, architecture, build, ship, run. Owned by a senior engineering lead with skin in the game.

  • Fixed-scope or T&M — we’ll recommend based on uncertainty
  • Engineering review before kickoff · we tell you if we’re wrong for the job
  • Production handoff includes runbooks, evals, on-call rotation training
  • Reference: Presight · Promvis — we’ve shipped our own
DURATION3–9 mo
TEAM3–8 eng
MODELFixed / T&M
SCOPE A PRODUCT →
SVC.03FLAGSHIP
FLAGSHIP PRODUCT · PRESIGHT.IO

AI Data Analyst

Presight is our AI Data Analyst — the only one designed to not hallucinate by construction. It runs on your verified metrics, in your VPC, on your warehouse. Production-deployed today.

  • Zero-hallucination architecture — answers anchored to certified definitions
  • Row + column-level security applied to every query, for every user
  • Connects to BigQuery, Snowflake, Redshift, SQL Server, Dremio
  • References: 1,000+ governed users (Metub) · zero-data-team SME launch in 1 week (Livwell)
DEPLOYVPC / on-prem
TIME-TO-VALUE3 days
HALLUCINATIONS0.00%
VISIT PRESIGHT.IO ↗
SVC.04METHOD

Agentic Development

Our internal methodology, codified. CLAUDE.md, multi-agent workflows, evals as a first-class artifact. We’ll embed it in your team.

  • Codified review of your repo, eval coverage, and agent harness
  • 2–6 week embedded engagement — we ship alongside, then hand off
  • Outputs: CLAUDE.md, agent runbooks, eval suites, PR conventions
  • Same methodology we use to ship Presight + Promvis — not a slide deck
DURATION2–6 wk
FORMATEmbedded
OUTPUTIn-repo
EMBED A POD →
SVC.05CLOUD

Cloud Transformation

AWS Advanced Partner with GenAI Competency. Greenfield, lift-and-shift, or refactor. Done by people who’ve carried the pager at scale.

  • AWS-first — can also deliver on Azure, GCP, on-prem hybrids
  • FinOps included — we don’t leave you with a $40k/mo dev cluster
  • Bedrock, SageMaker, ECS, Lambda; we keep current with the stack
  • Compliance: SOC 2, ISO 27001, SBV (Vietnam banking), HIPAA where required
PARTNERAWS · Advanced
COMPETENCYGenAI
DURATION1–6 mo
START AUDIT →
§ 02  /  CHOOSE

Embedded vs. dedicated — how to pick.

Most engagements start as one of these two shapes. Pick by control, not by price.

Embedded · SVC.01
Dedicated · SVC.02
You own
Repo, IP, roadmap, on-call. We follow.
Outcome. We own discovery → ops.
Best when
You have a tech lead and need senior bandwidth.
You need an AI product but don’t have an in-house lead.
Ramp
≤ 3 weeks
4–6 weeks (discovery)
Min commitment
6 weeks
3 months
Reporting
To your tech lead
To your exec sponsor
Eval / runbook handoff
— (you own)
Included
§ 03  /  FAQ

Things people actually ask.

Q.01Why hire AI engineers in Vietnam over India or Eastern Europe?+
Honest answer: rates land between Eastern Europe and India, but the senior bench is younger and AI-fluent by default — the average CoderPush engineer started using LLMs in production before they had five years of experience. We’re also smaller than the Indian giants, which means you talk to senior people, not account managers.
Q.02Do you sign data processing agreements? Where does code/data live?+
Yes — SCC + DPA standard for EU clients, BAA for HIPAA where applicable. Code lives in your repo, on your infra. Engineers work on company-issued machines with FDE and MDM. We never ship customer data offshore for training.
Q.03What happens if a CoderPush engineer rolls off mid-project?+
Two-week handover, including pairing with the replacement on live PRs. We carry the risk — you don’t pay during overlap. Our average tenure on a client engagement is 14 months.
Q.04Can you work with Claude / OpenAI / Bedrock / open-source models?+
All of the above, plus self-hosted Llama/Mistral/Qwen behind a VPC. Model choice is a function of latency, cost, and governance — not vendor preference. We’ll recommend based on the workload.
Q.05What does “AI-native” actually mean at CoderPush?+
Every engineer ships with Claude Code or equivalent in their loop. Every repo has a CLAUDE.md. Every PR has codified evals where the surface allows. It’s a habit, not a label.
▸ NOT SURE WHICH?

Tell us the problem. We’ll tell you the shape.

Free 30-minute call with a senior engineer. We’ll recommend embed vs. dedicated — or tell you you’re hiring the wrong team.

BOOK SCOPING CALL →
▸ SEE THE WORK

Read a real engineering write-up.

Our case studies are postmortems, not brochures. Architecture diagrams, trade-offs, named engineers.

SEE ALL WORK →