Vimarsh Patel
AI Software Engineer at White Collar Technologies, Inc
Developing agentic reconciliation systems and moving AI from 'vibe-coding' to production-grade reliability.
Intro
I'm Vimarsh Patel, an AI Software Engineer currently building agentic systems at White Collar Technologies. My career has been defined by a transition from full-stack development to the 'backend spine' of AI, including roles as a founding engineer and a consultant for Dow Jones. I specialize in moving AI beyond 'vibe-coding' into production-grade software, with a heavy focus on multi-agent workflows, data normalization, and MLOps. I am passionate about building reliable, deterministic systems that solve real-world problems in fintech and logistics. Whether it's automating complex three-way reconciliations or optimizing demand forecasting, I care about the data quality that makes AI actually work. I’m looking to connect with founders and engineers who value context over prompts and are interested in the practical deployment of agentic systems.
Networking
What I can offer
- ›End-to-end ownership of AI systems
- ›Orchestration and evaluation of multi-agent workflows
- ›Cloud deployment and MLOps expertise
- ›Fintech-specific AI domain knowledge
What I'm looking for
- ›Founders of early-stage startups
- ›AI engineers focused on production-grade systems
- ›Opportunities to own the 'backend spine' of AI products
Best fit for
Focus
Current interests
Core competencies
Background
Career
Progressed from software developer and team lead to specialized AI roles, serving as a founding engineer at two startups and consulting for Dow Jones before joining White Collar Technologies.
Education
MS in Computer Science from Illinois Institute of Technology (2022–2024); Bachelor’s in Computer Science from SRM IST Chennai (2018–2022); Entrepreneurship Studies at Entrepreneurship Development Institute of India (2017).
Achievements
- ›Automated a three-way reconciliation system across ADP, QuickBooks, and Outlook.
- ›Reduced manual data movement effort by 40% at Dow Jones.
- ›Boosted profit margins by 10% and forecast accuracy by 12% at Nastasi Foods.
- ›Reduced release cycles by 30% and improved uptime by 20% at Blackwins Tech Solutions.
- ›Led a 10-person development team to deliver 8 high-quality applications.
Opinions
- RAG is often overused; simpler Semantic Search is often superior for specific data shapes.
- Codebase context is 90% of the work; the prompt is only the last 10%.
- AI doesn't fix dirty data; it inherits it.
- LLMs should be removed from decisions that don't require language reasoning in favor of direct Python calls.
Personality
Communication style
Professional and analytical with a slightly contrarian edge, focused on debunking AI hype and emphasizing technical correctness.
Formality — 6/10
Vocabulary