SB-1456-AI Engineer

AI Engineer

Role Summary We are looking for AI Agent Architects to design and build production-grade agentic AI systems. This is a deeply hands-on engineering role centred on multi-agent orchestration, advanced context management, and large language model (LLM) integration, strong data structures and algorithms (DSA) skills, and hands-on ability with Python. You will work on the design of agent workflows and the context architecture that makes them reliable, accurate, and efficient, taking systems from prototype to production. The ideal candidate has a strong academic record, sharp problem-solving ability, and genuine enthusiasm for going deep on the agentic AI stack.

Key Responsibilities Agent Orchestration & Workflow

• Design and implement multi-agent workflows using LangGraph on Python with Pydantic structured output.
• Model complex, long-running processes as stateful, resumable graphs with branching, looping, retries, and durable checkpointing.
• Implement safe pause/resume and human-in-the-loop (HITL) checkpoints.

Context Engineering

• Engineer context management as a first-class subsystem — layered context, retrieval/indexing, and active working sets.
• Implement deterministic context selectors and filters, token-budgeted prompts, and summarisation/compaction of long histories.
• Design typed context schemas so each agent step receives precise, high-signal context.

LLM Integration & Retrieval

• Integrate LLM providers (e.g. Anthropic, OpenAI / Azure OpenAI, and self-hosted models) using robust prompt engineering, tool calling, and structured output.
• Wire in retrieval — vector search and embeddings — and code-intelligence techniques for working over large codebases.
• Contribute to model-routing logic that balances task type, risk, latency, and cost.

Quality, Evaluation & Governance

• Build evaluation and error-analysis loops; treat failures as feedback that improves reliability over time.
• Implement verification and validation patterns and deterministic gates for agent outputs.
• Ensure agent decisions and context are observable, auditable, and reproducible.

Collaboration

• Partner with platform/infrastructure engineers on deployment, inference, persistence, and durable execution.
• Contribute to engineering standards, design reviews, and code quality.

Required Technical Skills Domain

Skills & Technologies Must / Preferred

CS Fundamentals & DSA Data structures, algorithms, complexity analysis, strong problem-solving

Must Programming Python 3.10+ (async, typing); clean, idiomatic code

Must Agent Orchestration LangGraph — graphs/state machines, checkpointers, HITL interrupts

Must Context Engineering Layered context, selectors/filters, summarisation & compaction, token budgeting

Must Agentic AI Development Multi-agent design, tool calling, structured output, verification patterns

Must LLM Integration Anthropic & OpenAI / Azure OpenAI SDKs, prompt engineering Preferred

Data Modelling Pydantic v2, JSON Schema / typed contracts Preferred

Retrieval Vector stores (e.g. Qdrant / Azure AI Search), embeddings Preferred

Context Protocol Model Context Protocol (MCP) — resources/tools, Streamable HTTP Preferred

Multi-agent Frameworks CrewAI, Microsoft Agent Framework Preferred

Durable Workflows Temporal (long-running, resumable flows) Preferred

Inference vLLM awareness (paged attention, batching, quantisation), model routing Preferred

Qualifications & Certifications

• Strong academic record — B.Tech / B.E. / M.Tech / MCA in Computer Science or a related field from a reputable institution (or equivalent).
• Strong data structures, algorithms, and problem-solving skills — a competitive-programming track record (Codeforces / LeetCode / ICPC / similar) is a strong plus.
• Hands-on Python, plus exposure to LLM / agentic AI through academic projects, internships, or work — with clear eagerness to go deep on LangGraph and context engineering.

Preferred Certifications

• Microsoft Certified: Azure AI Engineer Associate
• Any recognised cloud certification (Azure / AWS / GCP) is a plus

Soft Skills & Cultural Fit

• Strong analytical mindset with a structured approach to design, debugging, and root-cause analysis.
• Clear written and verbal communication — able to explain agent and context design to technical and non-technical stakeholders.
• Comfortable with ambiguity and able to work independently with minimal supervision.
• Collaborative team player who contributes to shared standards, code reviews, and knowledge sharing.

What We Offer

• Deep, hands-on work with the modern agentic AI stack — LangGraph, MCP, and multi-agent systems.
• High ownership and influence over architecture from an early stage.
• Competitive compensation with a structured performance review process.
• Professional development support — certifications, conferences, and access to emerging tooling.
• Collaborative, transparent culture with clear growth pathways toward Staff / Principal engineering.

About Softobiz Softobiz is a technology firm. We build software and run dedicated offshore teams — Global Capability Centres — for a portfolio of international clients across multiple markets. We are looking for an Operations Manager to run the operations of the whole business: the delivery engine behind our GCCs, and the day-to-day running of the company itself. Just as important, we want someone who treats operations as something to be continuously sharpened — not just kept running. A core part of this role is reviewing how we work, redesigning processes that are manual or inefficient, automating them, and bringing AI deeply into how the operation runs. We are a technology company and we expect our operations to reflect that. Strategy and commercials sit with leadership; your job is to make the operation run, make it run better over time, report on it clearly, and escalate the right things at the right time. You will have an operations support resource handling day-to-day administrative and coordination tasks beneath you.

For more information about our solutions and organization, visit www.softobiz.com, Follow us on Twitter, Facebook, and LinkedIn.