Evals for Coding Agents

About the author behind Eval Basics

Yes, a Yahoo email in 2026. I love to hold on to old stuff — Yahoo Mail is one of them (right up there with my collection of USB flash drives I have no computer left to plug into).

About Me

I am Daya. I am a technology professional with more than 28 years of experience designing and delivering enterprise software solutions for some of the world's largest Insurance and Financial Services organizations. Throughout my career, I have helped organizations build, modernize, and scale mission-critical applications that power core business operations.

What makes my journey unique is the combination of expertise I have developed across Software Engineering, Data Engineering, and AI Engineering. Having worked extensively in all three disciplines, I bring a holistic perspective to solving complex technology problems—from building robust enterprise applications and large-scale data platforms to designing intelligent AI systems that augment software development.

My software engineering background includes legacy modernization, enterprise integration, API development, distributed systems, and cloud-native architectures. As a data engineer, I have designed and implemented large-scale data pipelines, data integration solutions, and analytics platforms that transform enterprise data into business value. Today, my focus is on AI Engineering, where I combine decades of software engineering principles with the latest advances in large language models, agentic AI, and intelligent automation.

An Agentic Engineering Mindset

What excites me most today is the emergence of Agentic AI and its potential to fundamentally transform software engineering.

I see coding agents as much more than code generators. The next generation of AI systems should behave like experienced software engineers—capable of understanding business requirements, analyzing large codebases, reasoning through architectural decisions, generating implementations, validating their work, identifying defects, and continuously improving through structured evaluation.

This perspective has shaped much of my recent work. I am particularly interested in building AI systems where multiple specialized agents collaborate to solve complex engineering problems while maintaining the reliability, traceability, and governance expected in enterprise software development.

A major focus of my work is developing rigorous evaluation (Eval) frameworks that measure the correctness, completeness, consistency, and trustworthiness of AI-generated software. As coding agents become increasingly capable, robust evaluation will become just as important as the models themselves.

Projects and Areas of Exploration

My recent projects explore the intersection of enterprise software engineering and artificial intelligence, including:

  • AI-powered code review systems that compare software implementations against business requirements and identify missing or incorrectly implemented functionality.
  • Evaluation (Eval) frameworks for coding agents that assess code quality, business rule compliance, reasoning accuracy, and overall reliability.
  • Multi-agent software engineering platforms using LangGraph and LangChain, where specialized agents collaborate on planning, coding, testing, reviewing, and documenting software.
  • AI-assisted modernization of COBOL-based enterprise applications into modern Java microservices while preserving decades of business logic.
  • Retrieval-Augmented Generation (RAG) architectures combining semantic search, vector databases, and structured knowledge to improve agent accuracy.
  • Intelligent developer tools that help engineers understand large codebases, trace business rules, and accelerate modernization initiatives.
  • AI-powered software development workflows that integrate automated testing, guardrails, evaluation pipelines, and human-in-the-loop review.
  • Data engineering solutions that support AI applications through high-quality data pipelines, retrieval systems, and scalable information architectures.

Across these projects, my objective remains consistent: to build AI systems that organizations can trust in real-world enterprise environments.

Passion for AI Engineering

Artificial Intelligence is redefining how software is designed, developed, and maintained. I am particularly fascinated by the convergence of software engineering discipline, data engineering foundations, and AI reasoning capabilities.

My interests extend well beyond prompt engineering. I enjoy exploring evaluation methodologies, agent orchestration, memory architectures, retrieval systems, deterministic validation, knowledge graphs, software architecture, developer productivity, and enterprise AI governance.

I believe the future belongs to engineers who can seamlessly combine software engineering, data engineering, and AI engineering to create intelligent systems that are both innovative and dependable. This multidisciplinary approach enables the development of AI solutions that are grounded in sound engineering principles while leveraging the latest advances in machine learning and large language models.

Why This Blog?

This blog is where I share practical insights drawn from nearly three decades of enterprise software development and my ongoing exploration of AI Engineering. My goal is to bridge the gap between traditional software engineering and the rapidly evolving world of AI coding agents.

Here you'll find articles on evaluation frameworks, coding agents, agent architectures, enterprise AI patterns, software modernization, RAG systems, LangChain and LangGraph, AI-assisted development, and lessons learned from building production-oriented AI solutions.

Whether you're a software engineer, data engineer, AI engineer, architect, engineering leader, or technology enthusiast, I hope these articles provide practical ideas that help you build more reliable, trustworthy, and intelligent software systems.

The future of software engineering will increasingly be powered by AI agents. Organizations that succeed will be those that combine strong engineering fundamentals with robust evaluation practices and high-quality data foundations. That is the journey I am passionate about exploring and sharing through this blog.