Generative AI Forward Deployed Engineer

Mactores is an AWS modernisation consultancy that describes itself as agent native: it runs customer engagements with an agent platform called Aedeon absorbing the repetitive discovery, dependency mapping and validation work, while its forward deployed engineers own architecture, judgement and the production cutover. This role is one of those engineers. You embed with a customer's engineering team, build and productionise AI agents, orchestration, retrieval pipelines, evaluation harnesses and observability against real customer data, and carry the delivery commitment personally on dates written into the contract. The posting is unusually direct about what it is not: not staff augmentation, not advisory. It is also direct about experience, saying roughly 3 to 10 years is indicative and that someone with 3 to 4 years of hands on agentic AI work typically outperforms a 12 year generalist on this work. Communication is listed as non negotiable.

What the posting requires:
- Excellent written and spoken English. The posting calls this non negotiable, because you present architecture to customer CTOs, write documents that hold up in audit, and defend decisions in the room.
- Production agentic AI systems you have actually shipped on AWS. The posting calls this the primary qualification and says to be ready to walk through what you shipped, the decisions you made and what broke. Proof of concepts and notebooks are explicitly not enough.
- Deep understanding of agentic architecture, including single agent versus multi agent systems, supervisor and orchestrator patterns, hierarchical topologies and planner and executor separation.
- Orchestration experience with LangGraph, Strands Agents, CrewAI or equivalent.
- Memory design: short term working memory and long term episodic and semantic stores, with vector and graph backed retrieval, and the production trade offs of each.
- Reflection and self correction patterns: critique loops, self evaluation, retry with feedback, and evaluation harnesses that catch failures before customers do.
- Tool use and function calling: schema design, tool selection reliability, error handling and agent to agent composition.
- RAG and retrieval pipelines: chunking, embedding, hybrid retrieval, reranking and grounding in customer data.
- Strong AWS production experience with Amazon Bedrock and AWS AI services, plus core services such as Lambda, API Gateway, DynamoDB, RDS or Aurora, Glue, EMR, Redshift and Kinesis.
- Solid software engineering fundamentals in Python and TypeScript, with CI/CD, infrastructure as code and test driven discipline.
- Roughly 3 to 10 years in engineering roles, with agentic AI or GenAI as your current day job.

Nice to have:
- Experience with data or application modernisation such as database migration, legacy refactoring or data platform builds.
- Delivery experience in financial services, healthcare and life sciences, internet and software, manufacturing, or telco, media, entertainment, gaming and sports.
- Model tuning and fine tuning.

What the work actually looks like:
- Deliver production agentic AI systems and AWS modernisation engagements on committed dates.
- Convert existing products and business processes into agents, exposing functionality as callable tools or replacing form and click interfaces with intent driven ones.
- Embed with customer engineering teams, run architecture sessions and align stakeholders from VP Engineering to CTO.
- Make agent decisions traceable and defensible so outputs survive internal audit and regulators, including HIPAA, PCI DSS and finance grade governance where the vertical demands it.

Location and working pattern: Mumbai, with the forward deployed model meaning you embed with customer teams.

Honest fit guidance: the deciding factor is whether you have shipped agentic systems into production, not how many years you have. The posting says so explicitly. If your agentic AI experience is demos, this application will not clear the first filter.