AI
AI / Next.js / Postgres
AI recruitment platform
A recruiter workspace that parses CVs, scores relevance, and makes shortlisting faster.
View project details ↗
Available for select projects
I'm Juned Soneji, a software engineer with 14 years of experience building dependable products and practical platforms for teams.
Based in
Surat, Gujarat01 / SELECTED WORK
A selection of products, experiments, and collaborations from the last few years.
02 / WHAT I DO
I enjoy bringing structure to difficult problems: understanding the business need, designing the right technical approach, and delivering software that remains maintainable as it grows.
I value practical architecture, clear communication, and software that is easy for the next engineer to understand.
Taking products from early idea through architecture, delivery, and iteration.
Exploring resilient delivery pipelines, cloud operations, and reliable AI systems in production.
Helping teams make practical decisions and turn complex work into momentum.
03 / LEADERSHIP & IMPACT
I care about outcomes that make teams faster, products stronger, and systems easier to operate.
Reduced the average CI/CD deployment cycle from roughly 30 minutes to 12 minutes.
Introduced stronger code review practices that improved quality and shared team knowledge.
Optimized database queries so year-long reporting could be generated in under one minute.
04 / TECHNICAL FOCUS
Select a focus area to see the tools and problems I work with.
From interface to infrastructure
Modern frontend frameworks, robust backend services, API design, microservices, and data layers that fit the product—not the other way around.
Reliable delivery
Practical cloud operations and CI/CD automation that reduce delivery friction while giving teams better visibility into the systems they own.
AI in production
Interested in the operational side of AI: evaluation, reliability, data flow, and integrating LLM capabilities into products people use every day.
Engineering excellence
I combine review culture, automated testing, performance work, and mentoring to build software that stays dependable over time.
05 / FROM THE LAB
SYSTEMS
DEVOPS
LLMOPS