Senior Software Engineer - AI Garage - HYD

ResponsibilitiesEngineering technical lead for the AI Garage team which responsible for developing AI/ML systems and Agentic solutions in HR domain Drive engineering excellence for the AI Garage team Provide technical leadership in regards to immediate and complex technical problems to leads of associated teams.Influence and Design and build scalable Agentic systems in HR domain. Collaborate with AI Garage Leads ( Eng Managers, Product Managers and AI strategist ) and stakeholders to understand their needs and translate them into technical solutions.Stay up-to-date with the latest trends in Agentic and AIML engineering.Required Skill :Master's degree in Computer Science, Software Engineering, or a related technical field, or equivalent practical experience8+ years of experience in software development, including proficiency in one or more programming languages, artificial intelligence, machine learning algorithms and tools, and Generative AI (Large Language Models, Multi-Model, etc.)Design and implement scalable AI solutions for Enterprises.Experience in Building and rolling out AI and agentic solutions for Global EnterprisesMulti-Agent Systems: Handle collaboration, memory persistence, and dynamic planning using frameworks like LangGraph, CrewAI, or AutoGen.Optimization & Evaluation: Track agent trajectories, manage cost/latency, and implement LLM-as-a-judge frameworks (e.g., Phoenix, LangSmith).Safety & Compliance: Prompt injection mitigation, sandboxed code execution, and human-in-the-loop (HITL) checkpoints.Experience with offline,online and RL based validationMetrics Selection: Knowing which evaluation metric matches specific business constraints (e.g., using Precision/Recall or Area Under the ROC Curve (AUC) rather than basic Accuracy on an imbalanced classification dataset).Good understanding of the Bias-Variance tradeoff, Overfitting vs. Underfitting, and practical techniques to fix them (L1/L2 Regularization, Dropout, Early Stopping). Dynamic Context Compaction: Skills in designing algorithms that programmatically truncate, summarize, or extract semantic milestones from a running conversation log so the agent doesn't suffer from "lost in the middle" phenomena during long-horizon tasks.Advanced Reasoning Frameworks: Implementing and customizing structural execution patterns like ReAct (Reason + Act), Tree of Thoughts (ToT), and Plan-and-Solve loops, giving models the cognitive scaffolding to break down vague goals into deterministic DAGs Evidence in runtime layers that scan user inputs for prompt injection or jailbreak attempts designed to hijack an agent’s system tools, alongside strictly isolating tool execution environmentsExperience with LoRA or SFT and reasoning