Skip to content
Search jobs inside your AI agent. Claude, ChatGPT, Claude Code, Codex, Gemini, Cursor and Grok
Connect

Mphasis

NewBe an early applicant

Software Engineer

Posted Updated

Summary

Senior (10+ yrs) engineer in Bangalore/Hyderabad building agentic GenAI systems on AWS — multi-agent workflows, MCP/RAG, LLM evaluation and LLMOps — orchestrated with Airflow, deployed via Kubernetes/Terraform CI/CD, and running on data platforms like Snowflake, Kafka, Pinecone, with model fine-tuning in PyTorch/Hugging Face.

Experience : 10+ yrs

Location : Bangalore / Hyderabad


AI & Agent SystemsAgent Design & Workflows:

Multi-agent systems, tool-using/self-learning agents, Model Context Protocol (MCP) for tool publishing, retrieval-augmented generation (RAG), Graph-RAG,

prompt engineering, model selection, and accuracy-per-cost routing.Evaluation & Guardrails: LLM evaluation harnesses (labeled sets, LLM-as-judge, RAGAS), guardrails, prompt-injection defense, audit trails, and rollback mechanisms.


Orchestration & AutomationWorkflow

Orchestration: Apache Airflow, AWS Step Functions, Amazon EventBridge for building durable, retryable, and idempotent AI pipelines.Task Management: Idempotent task design, retries, failure recovery, and backfills.


AI-Assisted Development

AI Coding Assistants: Claude, Cursor, Codex-style tools for scaffolding, refactoring, test generation, and code review


Cloud & DeploymentCloud Platforms:

AWS (Bedrock, S3, Glue, Lambda, EMR, Athena, CloudFormation, EventBridge), Kubernetes, Docker.Infrastructure as Code: Terraform (reusable modules), GitHub Actions, Jenkins, CI/CD, GitOps-style declarative deployments


Observability & LLMOps

  • Monitoring & Tracing: Datadog, Grafana, Amazon CloudWatch, LangSmith for distributed tracing, prompt versioning, token/cost and latency monitoring, alerting, and regression-eval gating.


Data Platforms & Streaming

  • Databases: Snowflake, PostgreSQL/pgvector, MongoDB, Amazon Redshift, Delta Lake, Apache Iceberg, Oracle, SQL Server, Teradata.
  • Event Streaming: Apache Kafka, Amazon Kinesis, Apache Flink.
  • Vector Search: Pinecone, embeddings, chunking, reranking.


Programming & Model Training

  • Languages & Frameworks: Python (pandas, PySpark, REST API), SQL, PL/SQL, Unix/Linux shell scripting, pytest.
  • Model Training/Serving: Fine-tuning (LoRA/PEFT, SFT, DPO/RLHF), distillation, quantization (GGUF/AWQ/GPTQ), self-hosted inference (vLLM, TGI), Hugging Face Transformers, PyTorch

Governance, Risk & Compliance

  • Controls & Validation: Regulatory reporting datasets, audit trails, RBAC, PII masking, data quality controls (Great Expectations, dbt tests), governed schemas/catalogs (AWS Glue Catalog, Erwin), dimensional/semantic data modeling.


Project Experience


  • End-to-End AI System: Built and deployed a custom agentic GenAI system (AROS) using AWS Bedrock, Airflow, Pinecone, and Kubernetes, with CI/CD via GitHub Actions and Terraform, monitored by Datadog and Grafana, and governed by audit trails and RBAC.
  • Data Platform Modernization: Consolidated multiple databases (Redshift, Oracle, Teradata, SQL Server) into a single governed source, orchestrated with Airflow and validated with Great Expectations


Skills

Apply

See also

Software Engineering jobs by country — openings, pay and top skills →