Turns Reddit, Twitter/X, and StockTwits noise into market signals — in real time.
About me
My name is Shreyas, and I'm a Computer Science and AI student at Northeastern University.
I've always been focused on building systems. That started with early experiments in hardware and simple programs, and evolved into software work in machine learning, backend systems, and real-time applications.
I've worked across research environments, public-sector IT, and product teams, with experience spanning XR systems, data pipelines, and scalable backend integrations. I also build independent projects, including ML-based recommendation systems and real-time analysis tools, with an emphasis on usability and performance.
I'm interested in systems that solve real problems through AI and infrastructure design. Outside of engineering, I spend time on badminton, movies, and gaming.
Experience
- Cloud Engineering & FinOps — designing and optimizing cloud infrastructure with a focus on cost visibility and resource efficiency across distributed systems.
- Architecting SentimentSignal's data ingestion layer — aggregating and normalizing posts from Reddit, Twitter/X, and StockTwits into a unified event stream processing thousands of data points per hour.
- Built an NLP classification pipeline that labels social posts as bullish, bearish, or neutral in real time, producing per-ticker sentiment scores across 50+ tracked equities.
- Designed the API schema and data model powering live sentiment dashboards with sub-second update latency and multi-source signal aggregation.
- Designed and built a haptic-feedback data visualization system in Unity — custom actuator integration lets users perceive multivariate dataset shifts through tactile response instead of visual cues alone.
- Engineered real-time IoT spatial sync pipelines connecting physical sensor arrays to a live Metaverse environment; achieved sub-50ms synchronization latency across distributed hardware.
- Led a 4-person team through full project lifecycle — from research design to working XR prototype — under Dr. Jin Ryong Kim's HCI lab at UTD.
- Produced technical documentation and multimedia assets for developer-facing product launches across LexisNexis's legal research platform.
- Designed motion graphics and visual campaigns in Adobe CC (After Effects, Photoshop) distributed to tens of thousands of legal professionals across the US.
- Built a backend integration service that synchronized shipping address records between Oracle EBS and Oracle Commerce Cloud, eliminating a class of FedEx delivery failures on county procurement orders.
- Reverse-engineered field mappings between two Oracle enterprise systems with no existing API contract; implemented reconciliation logic to detect and auto-correct address mismatches pre-shipment.
- Wrote PL/SQL validation and transformation scripts standardizing address data across thousands of county records — reducing failed deliveries by ~30%.
Projects
VR curriculum for students who learn better by touching things. Anatomy, astronomy, geometry — on Meta Quest.
Two filtering models, one engine. Ranked movie recommendations in under 100ms.
Stopped a recurring class of failed government deliveries. No API existed — so I reverse-engineered one.
Recommends what to wear based on today's actual weather. Full-stack, your wardrobe, live forecast data.
Real-time Azure telemetry pipeline. Event-driven architecture built the way production monitoring systems actually work.
Skills & tools
Let's build something.
Based in Atlanta & Boston
Let's connect if you're working on something cool — or if you just want to talk tech, opportunities, or anything else.