Software Engineer 3 - GenAI

Bidgely, Inc; India; In loco; INR 4.500.000 - 7.000.000; Tempo pieno

Descrizione del lavoro

Bidgely, Inc. seeks a senior software engineer to build and scale Energy Assistant, a GenAI-powered agent integrated in utility portals and apps. You will own major components end-to-end, shape architecture, and raise engineering standards in a hands-on role.

You will work closely with GenAI engineering, platform, QA/SDET, and product teams to deliver robust, low-latency services and secure, tenant-isolated data handling in a multi-tenant SaaS environment.

Competenze

• 6-10 years of software engineering experience, including 1-2 years building and shipping LLM or GenAI features in production.
• Strong Python skills and good working knowledge of Java.
• Hands-on experience with LangChain, LangGraph, or similar agent orchestration frameworks.
• Experience with LLM APIs (OpenAI, Anthropic, Amazon Bedrock), prompt design, tool/function calling, and structured outputs.
• Experience building RAG systems with embeddings and vector stores (e.g., pgvector).
• Strong backend and API design skills, with experience building scalable, low-latency services.
• Experience with AWS, containers (EKS), PostgreSQL, Redis, and CI/CD.
• Experience with evals, LLM-as-judge, or AI observability tools such as LangSmith.
• Strong system design, analytical, and communication skills.

Mansioni

• Design and build production multi-agent systems: supervisor/router and specialist agents, LangGraph workflows, tool calling, conversation memory, and checkpointing.
• Build typed tool contracts and data-access layers over Bidgely APIs to ensure the LLM explains results instead of generating numbers.
• Develop RAG pipelines for utility tariff and knowledge retrieval, with tenant-isolated embeddings and retrieval.
• Manage latency, reliability, and cost: caching, prewarming, retries, circuit breakers, and model routing across small and large model tiers.
• Implement guardrails for prompt injection, scope control, authorization, tenant isolation, and data management.
• Build observability and evaluation hooks with LangSmith, and collaborate with SDETs on evaluation datasets, regression suites, and release criteria.
• Debug agent failures in production using traces, logs, and metrics, and drive root-cause fixes.
• Support streaming (WebSocket/SSE) and widget integration with front-end teams.
• Lead design and code reviews, set engineering standards, and mentor SDE1 and SDE2 engineers.
• Communicate designs, trade-offs, and progress to engineering, product, and leadership.

Conoscenze

Python

Java

LangChain

LangGraph

LangSmith

LLM/GenAI features

APIs design

System design

Communication

Strumenti

OpenAI API

Anthropic

Amazon Bedrock

pgvector

REST APIs

WebSocket/SSE

Kubernetes (EKS)

PostgreSQL

Redis

CI/CD

Descrizione del lavoro

About Bidgely

Bidgely is a Silicon Valley AI company transforming how utilities and their customers understand and manage energy. The UtilityAI platform converts smart meter data into appliance-level insights, while Agentic CX brings GenAI assistants to utility chat, voice IVR, and customer service. Leading utilities across North America, Europe, and Asia rely on Bidgely. Our Bengaluru engineering center is a certified Great Place to Work.

The Role

We are seeking a senior software engineer to build and scale Energy Assistant, our GenAI assistant embedded in utility web portals and mobile apps. Energy Assistant answers customers' questions about usage, bills, appliances, rate plans, and savings through a multi-agent system built on LangGraph, leveraging Bidgely's energy data with tool calls and deterministic calculations. You will own major components end to end, help shape the architecture, and elevate the engineering standards of the team. This is a hands-on role working closely with GenAI engineering, platform, QA/SDET, and product teams.

Key Responsibilities

• Design and build production multi-agent systems: supervisor/router and specialist agents, LangGraph workflows, tool calling, conversation memory, and checkpointing.
• Build typed tool contracts and data-access layers over Bidgely APIs to ensure the LLM explains results instead of generating numbers.
• Develop RAG pipelines for utility tariff and knowledge retrieval, with tenant-isolated embeddings and retrieval.
• Manage latency, reliability, and cost: caching, prewarming, retries, circuit breakers, and model routing across small and large model tiers.
• Implement guardrails for prompt injection, scope control, authorization, tenant isolation, and data management.
• Build observability and evaluation hooks with LangSmith, and collaborate with SDETs on evaluation datasets, regression suites, and release criteria.
• Debug agent failures in production using traces, logs, and metrics, and drive root-cause fixes.
• Support streaming (WebSocket/SSE) and widget integration with front-end teams.
• Lead design and code reviews, set engineering standards, and mentor SDE1 and SDE2 engineers.
• Communicate designs, trade-offs, and progress to engineering, product, and leadership.

Requirements

• 6-10 years of software engineering experience, including at least 1-2 years building and shipping LLM or GenAI features in production.
• Strong Python skills and good working knowledge of Java.
• Hands-on experience with LangChain, LangGraph, or similar agent orchestration frameworks.
• Experience with LLM APIs (OpenAI, Anthropic, Amazon Bedrock), prompt design, tool/function calling, and structured outputs.
• Experience building RAG systems with embeddings and vector stores (e.g., pgvector).
• Strong backend and API design skills, with experience building scalable, low-latency services.
• Experience with AWS, containers (EKS), PostgreSQL, Redis, and CI/CD.
• Experience with evals, LLM-as-judge, or AI observability tools such as LangSmith.
• Strong system design, analytical, and communication skills.

Nice to Have

• Experience with time-series, IoT, or smart meter data, or the energy and utilities domain.
• Experience with voice or IVR GenAI products.
• Experience with multi-tenant SaaS solutions, security, and data-privacy enhancements for LLM applications.

Tech You Will Work With

Python, Java, LangChain, LangGraph, LangSmith, OpenAI, Anthropic, Amazon Bedrock, RAG, pgvector, tool calling, agent guardrails, REST APIs, WebSocket/SSE, PostgreSQL (RDS), Redis, AWS (EKS), ELK, and CI/CD.