ML Engineer I

UST; Mumbai; Presencial; INR 2.500.000 - 4.500.000; Jornada completa

Descripción de la vacante

UST in Mumbai (Nesco) is seeking an AI Forward Engineer – Apollo ISG to design and deploy production Gen AI solutions within our operations platforms.

You will own pipelines, accuracy controls, and end-to-end delivery, collaborating with Loan Ops and Investment Ops to automate processes.

Proficiency in Python and C#/.NET or TypeScript, SQL, Azure, and experience with LLM/RAG and prompt engineering is required.

Formación

• Strong Python, plus one of C#/.NET or TypeScript.
• Solid SQL knowledge.
• Hands-on production experience with LLM applications, RAG, and structured extraction from unstructured documents and emails.

Responsabilidades

• Build production Gen AI and agentic components inside our operations platforms, primarily document and email reading, extraction, classification and enrichment feeding downstream posting into PAM and other systems.
• Design prompt and configuration driven pipelines so a reader built for one workflow can be extended to another asset class or servicer without a code rewrite.
• Own the accuracy and control layer, deterministic validation, exception routing, evaluation harnesses and regression testing of model outputs.
• Work directly with Loan Ops and Investment Ops users to convert manual processes into straight-through flows, including requirement shaping, not just delivery.
• Ship end to end, API, workflow orchestration and the front end needed for Ops to review and act on exceptions.
• Strong Python, plus one of C#/.NET or TypeScript. Solid SQL.
• Hands-on production experience with LLM applications, RAG, structured extraction from unstructured documents and emails, prompt engineering and output validation. Agent frameworks and tool/function calling.
• Cloud native delivery on Azure/AKS, containers, CI/CD.
• Workflow orchestration experience, Temporal preferred.
• Ability to work directly with business users and translate an operational process into a system design.

Conocimientos

Python

C# / .NET

TypeScript

SQL

LLM applications

Prompt engineering

Agent frameworks

Tool/function calling

Azure / AKS

CI / CD

Temporal

React frontend

Snowflake

Okta

Herramientas

Azure / AKS

Descripción del empleo

Role Description

Role: AI Forward Engineer – Apollo ISG Nesco Experience: 4 to 8 years. Location: Mumbai, Nesco. What the role does

• Build production Gen AI and agentic components inside our operations platforms, primarily document and email reading, extraction, classification and enrichment feeding downstream posting into PAM and other systems.
• Design prompt and configuration driven pipelines so a reader built for one workflow can be extended to another asset class or servicer without a code rewrite.
• Own the accuracy and control layer, deterministic validation, exception routing, evaluation harnesses and regression testing of model outputs.
• Work directly with Loan Ops and Investment Ops users to convert manual processes into straight-through flows, including requirement shaping, not just delivery.
• Ship end to end, API, workflow orchestration and the front end needed for Ops to review and act on exceptions. Must have
• Strong Python, plus one of C#/.NET or TypeScript. Solid SQL.
• Hands-on production experience with LLM applications, RAG, structured extraction from unstructured documents and emails, prompt engineering and output validation. Agent frameworks and tool/function calling.
• Cloud native delivery on Azure/AKS, containers, CI/CD.
• Workflow orchestration experience, Temporal preferred.
• Ability to work directly with business users and translate an operational process into a system design. Good to have
• Financial services operations background, loan servicing, remittance, custody, accounting or investment operations.
• Snowflake or similar data platform, React front end exposure, Okta based access models.
• Evaluation and observability tooling for AI systems.
• What we are screening for
• Evidence of AI features actually running in production with users, not proof of concept work.
• Judgement on where AI should and should not be used, and comfort putting deterministic controls around model output.
• Speed and ownership. This team ships small increments directly to Ops every sprint.