About Trangile

Trangile builds enterprise AI that runs where our clients' data lives. Many of our deployments sit on client servers, some on networks with no internet access at all; others run inside the client's own

AWS, Azure or Google Cloud tenancy. Our position is simple: Own your AI. Own your data. Own your costs.

We work with enterprises across India, the UAE and the wider Gulf, Southeast Asia and Egypt,

mostly in retail, supply chain, procurement and finance operations. Our products include:

About the Role

You will own the technical design of AI solutions from the first client conversation through to production. That means joining scoping calls and POCs, deciding the architecture, and being clear about where an LLM genuinely helps and where plain rules do the job better. Then you build it and deploy it: on client hardware, in the client's cloud, or across both. You will also lead a team of engineers and set the standard for how we build.

Key Responsibilities

- Own solution architecture end to end: data flows, model selection, deterministic logic, agent design, deployment topology and hardware sizing.
- Decide where deterministic, rule-based logic should carry the load and where LLM reasoning adds value, and design the validation, guardrails and fallbacks between the two.
- Build multi-agent and RAG systems using LangGraph, LangChain, Google ADK, MCP or equivalent, and integrate them with client ERPs, databases and document stores.
- Deploy and operate models in both on-premises and cloud environments: self-hosted open-

weight models on client GPUs, including air-gapped networks, as well as managed AI services and Kubernetes on AWS,



Azure or Google Cloud.
- Fine-tune open-weight models (LoRA/QLoRA) where retrieval and prompting are not enough,

and recognise when fine-tuning is not worth the cost.
- Build forecasting and tabular ML models for allocation, replenishment and cost estimation use cases.
- Define evaluation for every system you ship: test sets built from client data, accuracy and hallucination metrics, and regression tests for prompts, models and agents.
- Support pre-sales with technical input on POCs, demos, solution scoping, effort estimates and infrastructure sizing, and run technical workshops with client teams.
- Make sure deployments meet data protection and residency requirements, including India's

DPDP Act, UAE and Saudi PDPL, and client-specific security policies.
- Mentor junior engineers, lead code reviews, and set engineering standards for testing,

documentation and CI/CD.
- Monitor production systems, diagnose failures and ship fixes.

Required Experience and Skills

- 6+ years in software engineering, including ML or AI engineering, with systems you have taken to production and supported there.
- Strong Python and solid backend engineering: APIs, services, asynchronous processing and testing.
- Hands-on experience building LLM applications, including RAG, agents,



tool use and structured output, using frameworks such as LangGraph, LangChain or Google ADK.
- Experience self-hosting open-weight models (Llama, Qwen, Mistral, Gemma or similar) with vLLM, Ollama, TGI or equivalent, including quantisation and GPU capacity planning.
- Experience deploying ML or LLM workloads on at least one major cloud platform (AWS, Azure or Google Cloud), for example through SageMaker, Bedrock, Azure ML, Azure OpenAI,

Vertex AI, or containerised on managed Kubernetes.
- Experience fine-tuning models, ideally open-weight LLMs using LoRA or QLoRA.
- Solid SQL, and working experience with at least one NoSQL store such as MongoDB.
- Grounding in classical ML, including forecasting, gradient boosting and evaluation metrics,

and judgement about when these beat an LLM.
- A track record of building evaluation into AI systems, not just demos.
- Docker, Kubernetes, Git and CI/CD.
- The ability to explain a technical trade-off to a client's finance or operations head without jargon, and to write clear design documents.

Preferred

- Deployments in air-gapped or highly restricted network environments.
- Building MCP servers.
- Document AI: OCR, layout parsing and table extraction from scanned PDFs.
- Knowledge graphs or graph databases.
- Spark or similar tools for large datasets.
- TypeScript, Node.js and React/Next.js for building demo interfaces.
- Domain experience in retail, supply chain, procurement or finance operations.
- Client-facing consulting or pre-sales experience.
- Open-source contributions or publications.

📌 Lead - AI & Machine Learning (Noida)
🏢 Trangile Services
📍 Noida