Two significant milestones happened for Yash Kamlesh Shah on May 20: he officially graduated with his M.S. in Data Science from the Ying Wu College of Computing; and his startup, Avarieux, was publicly announced, with a waitlist that opened once the company came out of stealth.
Avarieux is an AI research platform for self-directed investors and financial advisors, built on a single design principle: structural honesty. Every claim the system surfaces is cited to a primary source, and every number is verified against that source before it reaches the user.
Current AI models are known to generate inaccuracies or “hallucinations” that appear confident but are entirely false or incorrect. With high stakes domains where structural honesty is critical, being “usually right” isn’t good enough, according to Shah.
“Every claim is cited, every number is verified against the source,” he said. "If the system can't ground a number in a primary source, it doesn't show you the number."
The idea behind Avarieux started with a simple observation: a politician's disclosure, a quiet regulatory filing, an event halfway around the world — small public signals ripple outward and move markets, often days before the news catches up. Most tools tell you what the market did. Shah wanted to build the one that surfaces the public events themselves, raw and cited, and lets the investor draw their own conclusions.
"Every public event is a ripple," he said. "We show you the ripples, with their sources. We never tell you what to buy — the call is yours."
As a dually minted graduate and CEO of a bourgeoning entrepreneurial venture, Shah shared some insight on his journey from international student to founder, and why NJIT’s Ying Wu College of Computing was a “deliberate” choice.
Why did you choose NJIT-Ying Wu College of Computing?
I came to the U.S. from India in fall 2024 specifically to study data science at YWCC, and the choice was deliberate. The program had three things I wasn't finding elsewhere at the same combination of price and depth: serious technical rigor in machine learning and information systems, faculty actively researching in the spaces I cared about, and a location in the NYC metro that put me near the financial-services and AI-engineering work I knew I'd want to engage with after graduation.
The other thing that drew me was the practitioner orientation of YWCC. A lot of data science programs trend either toward pure theory or toward narrow industry tooling. YWCC's curriculum balanced both — I came out understanding both the mathematics underneath modern AI and the engineering reality of shipping it. That balance turned out to be exactly what I needed for what I'm doing now.
What were some highlights of your experience while at the university and/or the college?
Three highlights stand out.
First, the technical depth of the data science coursework — statistical methods, machine learning architectures, and applied data engineering. I came in with a strong programming background but limited formal grounding in ML theory, and YWCC closed that gap rigorously. The verification approach I now use at Avarieux builds directly on principles I strengthened in YWCC coursework.
Second, the faculty access. I was able to engage directly with professors on questions that went beyond the classroom — about applied research, about industry direction, about what to read next. That access isn't guaranteed at every program, and at YWCC it was real.
Third, the NJIT Center for Student Entrepreneurship. Dr. Kathleen Naasz and the broader entrepreneurship community opened doors to the founder side of my path at exactly the right time. Sitting in that ecosystem changed how I thought about what I could build, not just what I could study. NJIT having such an active ecosystem for graduate students is a quiet but powerful feature of the institution.
Why might you recommend attending NJIT-YWCC to someone looking to study computing?
Three reasons, in order of importance.
Rigor without inflation. YWCC trains you for real technical work, not just credentials. The M.S. Data Science program goes deep enough that what you learn translates into work that ships. I'm using what I learned at YWCC in my company right now — that's the test that matters.
Location is leverage. Being twelve miles from Manhattan means you're inside the AI-finance and AI-engineering ecosystem shaping the next decade. I started building my company while still in the program because the geography made it possible.
The community is real, not performative. The faculty engage. The entrepreneurship center connects you to founders and mentors. The international student network is strong and genuinely supportive. I came here not knowing anyone in the country, and I'm leaving with a network that's already changing what's possible for me.
If you're trying to build a serious technical career in data, AI, or computing — especially in the Northeast — YWCC is the right call.
“I'm using what I learned at YWCC in my company right now — that's the test that matters.”
What are your responsibilities at Avarieux and what technology do you work with?
I'm responsible for everything — engineering, product, fundraising, and operations. The technical core is what I built personally: a multi-source data infrastructure layer that ingests regulatory filings, government economic data, prediction markets, news flow, and alternative datasets; a grounding verifier that audits every claim the AI makes against the underlying source before it reaches the user; and a citation archive that gives every analysis a timestamped, citable record.
The stack is Python and TypeScript across the backend, with the agent layer built on the Model Context Protocol (MCP) — the open standard Anthropic released in late 2024 that lets large language models interact with real tools and real data. I've authored six open-source MCP servers spanning financial filings, prediction markets and alternative data, real-time audio analysis, anti-detection web browsing, and neuroscience research tooling. I'm also a merged contributor to Anthropic's official modelcontextprotocol/servers repository, and my servers have been picked up and listed across MCP registries and directories, including PulseMCP and Claude Marketplaces.
What is it about this job that is most rewarding?
Two things. Building something that takes structural integrity seriously, at a moment when most AI products don't, is meaningful. Markets are downstream of the world, and if the tools people use to understand markets are inventing their own facts, that's a real problem worth solving.
Second, autonomy and velocity. As a founder, I make every decision about what to build, how to ship it, and who to bring along. There's nothing else like it.
Any tips for rising graduates seeking to transition into industry?
I didn't get my current job — I built it.
Six months ago, I was rejected at the offer-letter stage of a role I'd interviewed through every round for. The final answer was a years-of-experience requirement I couldn't meet. That rejection forced a different path — and it turned out to be the best thing that could have happened. The greatest gifts really are in the detours.
What followed was a year of disciplined building. I joined Papex as a founding engineer, which gave me an inside view of what running a startup actually looks like. In parallel, I went deep into open-source AI. Avarieux grew out of all of that.
A few things I'd tell graduating YWCC students who are wondering how to make this kind of path work for them:
- Public work compounds. A LinkedIn post nobody reads doesn't compound. An open-source repo with real users does. Spend your time on artifacts that exist publicly and improve over time.
- Build something specific, not generic. I didn't start by trying to build a company. I started by building a tool that solved a specific problem I had — and that tool became the foundation for what came next. Specificity is the path; abstraction is the trap.
- Use the NJIT ecosystem. The Center for Student Entrepreneurship, the faculty, the alumni network, the international student community — these are real resources, not theoretical ones. Show up to office hours. Ask for introductions. Most people are willing to help if you ask clearly.
- The rejection isn't the end of the story. It's the first chapter. The interview process is a feedback loop, not a judgment. Use what you learn and keep moving.
