MT

AI Architect

Accepting applications

Mirafra Technologies · Bengaluru, Karnataka, India

Full-Time Mid_senior AIARMCalibreDFTInnovus
Estimated market salary
₹45-81 LPA

This is a SiliconBoard market estimate, not an employer-posted salary.

Posted
4d ago
Category
Eda
Experience
Mid_senior
Country
India
Role Overview

We are seeking an AI Architect to lead the design and integration of AI/ML capabilities across our semiconductor engineering workflows and service offerings. The role involves bridging deep AI/ML systems expertise with hands-on understanding of the chip design lifecycle (RTL-to-GDSII, DFT, verification, sign-off). The ideal candidate will architect AI-augmented EDA workflows, build internal AI tooling, and shape AI-differentiated service lines for our clients.
As agentic AI reshapes how engineering work gets done, this role will pioneer how autonomous and semi-autonomous agents are embedded into our design and verification flows — moving beyond AI-as-assistant toward AI-as-collaborator.
Key Responsibilities
Define the AI architecture strategy spanning data pipelines, model selection, and deployment for semiconductor design use cases.
Architect and prototype AI/ML solutions that augment backend design, STA, DFT, and physical verification flows (e.g., PPA optimization, automated DFT insertion, log/report analysis, predictive timing closure).
Integrate LLMs and ML models into existing EDA toolchains (Innovus, Tempus, PrimeTime, Calibre, etc.) and Linux/TCL/Python scripting environments.
Architect agentic AI workflows — agents that plan, execute, and iterate across multi-step design tasks (debug triage, ECO loop automation, constraint generation, regression analysis) with appropriate human-in-the-loop checkpoints.
Design multi-agent systems that orchestrate specialized agents across the RTL-to-GDSII flow, integrating tool calls to EDA engines, scripts, and internal knowledge bases.
Build robust agent infrastructure: tool/function-calling interfaces, context and memory management, state handling, and guardrails for reliability in engineering-critical applications.
Evaluate and benchmark foundation models, fine-tuning approaches, RAG pipelines, and agent orchestration patterns.
Establish an agent control plane — policy, approval gates, observability, and audit trails — before autonomous agents touch production systems.
Partner with delivery teams and clients to translate AI capabilities into productized service offerings.
Establish MLOps practices, model governance, and scalable infrastructure (on-prem/cloud/hybrid).
Mentor engineers on AI/ML adoption and build the technical foundation for an AI Center of Excellence.
Required Qualifications
Bachelor's/Master's in CS, ECE, EE, or related field; advanced degree preferred.
10+ years in software/AI engineering, with 3+ years architecting production AI/ML systems.
Demonstrated experience with deep learning frameworks (PyTorch, TensorFlow), LLM application development, and modern ML infrastructure.
Hands-on experience building production agentic systems with one or more current frameworks — e.g., LangGraph (graph-based, strong for stateful production workflows), CrewAI (role-based multi-agent crews), AutoGen/AG2 (conversational multi-agent), or the Claude Agent SDK / OpenAI Agents SDK for model-native agents.
Working knowledge of the Model Context Protocol (MCP) for connecting agents to tools and data sources, and familiarity with agent-to-agent coordination patterns.
Strong command of tool-use/function-calling patterns, ReAct-style planning loops, RAG, and agent evaluation/observability.
Strong proficiency in Python; familiarity with data engineering and MLOps tooling.
Working knowledge of the semiconductor/chip design flow — RTL, synthesis, P&R, STA, DFT, and/or physical verification.
Preferred Qualifications
Hands-on background in VLSI/backend design or EDA tool development.
Experience with EDA-vendor AI offerings, AI-assisted design automation, or agentic coding tools.
Exposure to advanced nodes (2nm, 3nm, 5nm…..22nm) and ARM-based SoC programs.
Familiarity with agent reliability engineering — failure recovery, retries, runaway-tool-call prevention, and audit readiness.
Track record of building AI products or service lines in an engineering services context.
Show more Show less