15 Nov
|
Ciklum
|
Hyderabad
Job highlights
- Expertise in building autonomous AI agents using LangGraph, AutoGen, and LLM hybrids with strong skills in AI-native workflows and legacy system modernization
- Design, build, debug, deploy, and maintain multi-step AI workflows and autonomous agents; automate legacy business logic testing; perform prompt engineering and rapid prototyping; ensure agent reliability and error handling
You are a full-stack agent engineer. You can take a complex business requirement and build the multi-step AI workflow that solves it, end to end, without someone breaking it down for you first. 1. Core DNA
- AI-Native Workflow. You use AI to accelerate development, testing and documentation as a default, not an
experiment. You have opinions about which tools earn their place.
- Agentic Implementation. You build, debug and deploy autonomous agents (LangGraph, AutoGen, Playwright-
LLM hybrids) that navigate complex ERP schemas and long-running business logic.
- Legacy-to-Modern Bridge. You treat 20-year-old systems as puzzles for AI-driven modernization, not as
someone elses problem.
- Evidence Over Confidence. You dont ship an agent because the demo went well. You ship it because you can
show what it does on the edge cases and you know your own error rate rather than claiming there isnt one. 2. What You Own
Your agent. Design, build, debug, deploy, and keep running. Including the unglamorous parts: retries, state recovery,
timeouts, what happens when the model is down.
The evidence it works. Before you refactor an undocumented module, you capture what it currently does characterization tests, real inputs, real outputs. After, you show the behaviour matches. Your Builder 3 sets the standard; meeting it is your job, not theirs.
Your blast radius. You know exactly what your agent is permitted to touch, and you build so a wrong answer costs a review comment rather than a bad ledger entry.
3. Your Elective Specialty (lean into one)
- The Agentic Architect (Systems) — you don’t just build APIs, you build tool-calling interfaces and state
management for agentic graphs that survive long-running business processes.
- The UX/Interaction Builder (Product) — the chatbox is just the beginning. You build generative UIs and
browser-use agents that navigate existing portals to get work done.
- The Data & Context Engineer (Data) — master of the RAG pipeline. Vector stores, chunking strategy, and
knowledge graphs that teach the AI how a Purchase Order relates to a GL entry.
- The Reliability & Eval Engineer (QA) — you move the team from prompt vibes to prompt engineering. DSPy
for programmatic optimization, red-teaming agents to pressure-test financial logic.
- The Infra & Agent-Ops Builder (DevOps) — you build the factory where agents live: tracing, logging and
monitoring the thoughts of the AI.
- The Domain Logic Translator (Business) — you take 50 pages of accounting regulation and turn it into
structured system prompts and agentic constraints.
What You’ll Actually Do
- Build execution agents. Move beyond RAG into action — agents that navigate an ERP schema, find a data
discrepancy and propose the fix.
- Automate the untestable. Wrap legacy business logic in contemporary tests and documentation using AI, starting
with the modules everyone else has avoided.
- Prove parity. Capture what the old code does before you replace it, and show the new path matches — on the
boring cases and the fiscal-calendar, multi-currency, partial-shipment ones.
- Prompt engineering and evals. Use frameworks like DSPy to move from vibes-based prompting to
programmatic, optimized instructions.
- Rapid prototyping. Turn a “what if” into a working AI-driven feature in days, not months.
- Fix your own agents in production. Read the traces, find why it went sideways, make it not happen again.
The Stack
- Python, SQL (Postgres/SQL Server), TypeScript, OpenAI/Anthropic, agentic coding tools, LangChain/LangGraph, DSPy,
- pgvector/Qdrant, dbt, Apache Hop/Airflow, Docker, Playwright, Weights & Biases, OpenTelemetry, eval and tracing
- platforms (LangSmith / Langfuse / Braintrust or equivalent). And any other tool that helps us achieve greatness.
📌 Senior Artificial Intelligence Engineer (Hyderabad)
🏢 Ciklum
📍 Hyderabad