Portfolio
CS × Math through a philosophizing lens
// work
Developed a constraint-rich portfolio optimizer that constructs equity portfolios under realistic investment mandates, including sector exposure limits, ESG/liquidity filters, beta ranges, volatility ceilings, cardinality requirements, and weight bounds. Modeled portfolio selection using Boolean decision variables and arithmetic constraints, framing portfolio construction as an SMT-style feasibility and optimization problem. Built a 50-asset large-cap universe and interactive dashboard that outputs selected holdings, allocation weights, share counts, risk metrics, sector exposure, and exclusion explanations. Implemented static GitHub Pages demo mode and local FastAPI solver mode, separating portfolio presentation from backend solver execution.
Developed an interactive web-based logic visualization platform for constructing and evaluating propositional formulas and logical circuits in real time. Designed dynamic visual representations of logical operators and truth-functional relationships, enabling intuitive exploration of formal reasoning processes. Implemented clientside parsing, state management, and live circuit evaluation to provide immediate feedback as users modified logical expressions. Engineered an educational interface that bridges symbolic logic and computational visualization, making abstract logical structures accessible through interactive graphics.
Developed an interactive platform for semantic authoring & formal reasoning, transforming mathematical prose into structured definitions, theorems, proofs, & dependency graphs. Implemented document parsing, logical diagnostics, & formal-export pipelines supporting knowledge representation & proof- assistant workflows. Engineered a React/TypeScript application using Vite & GitHub Pages with dynamic visualization of logical relationships & AI-oriented formalization infrastructure. Built interactive visualizations of logical dependencies & conceptual relationships, allowing users to explore the structure of mathematical arguments & formal theories through dynamic graph-based interfaces.
Designed & built a robust simulation framework that models the probabilistic behavior & wave function of a single quantum particle in a bounded one-dimensional potential. Achieved realistic quantum state representations & validated results against standard quantum mechanical principles.
Developed a series of AI models using Computer Vision, TensorFlow, Keras & HuggingFace transformers for trash identification across complex terrains as a part of a group. Collaborated with a research team to analyze & map trash accumulation on Allen Island, contributing to environmental studies.
Dynamic topic classification using guided machine learning on scientific literature spanning 5+ decades, combined with geographic data analysis to trace the evolution of ice-core research across time and institutions.
// experience / activities
Research, analytics, and teaching — the working record behind the research agenda: algorithms that became a publication, machine learning in climate science, and four years of making logic click for other people.
Designed scalable Python data-collection frameworks tracking user acquisition, engagement, retention, and platform performance for a nonprofit serving stakeholders in 190+ countries — then built the interactive dashboards and KPI systems that turn that telemetry into decisions, powering impact measurement, A/B experimentation, and strategy across product, research, and operations.
Investigated dynamic pricing for the online set cover problem in the client–server model, where a server can only post prices rather than assign resources. Helped characterize which online algorithms can be mimicked by pricing schemes and analyze competitive efficiency — work that became the co-authored paper Dynamic Pricing Algorithms for Online Set Cover (arXiv:2409.15094).
Mapped five decades of ice-core science through comprehensive bibliometric analysis, building guided machine-learning models in PyTorch and TensorFlow for dynamic topic classification — tracing how the field's questions, methods, and institutions evolved across time and geography.
Teach across seven courses spanning computer science, mathematics, and philosophy — CS333, CS252, CS231, CS152, MA130, MA120, and PL151. Grading, help sessions, and the patient craft of making formal reasoning click for someone else, one explanation at a time.
activities
Intensive program spanning epistemic logic and topology, decision theory and the foundations of AI agents, and the path from “gavagai” to a full grammar.
Collaborative machine-learning practicum — applying ML models to real datasets and building hands-on fluency with data-driven AI workflows.
// research
We consider dynamic pricing algorithms as applied to the online set cover problem. In the dynamic pricing framework, we assume the standard client server model with the additional constraint that the server can only place prices over the resources they maintain, rather than authoritatively assign them. In response, incoming clients choose the resource which minimizes their disutility when taking into account these additional prices. Our main contributions are the categorization of online algorithms which can be mimicked via dynamic pricing algorithms and the identification of a strongly competitive deterministic algorithm with respect to the frequency parameter of the online set cover input.
Read on arXiv ↗// me
// @pramithxs
// watching
89 shows
// celestial navigation
Every section of this site is a star. Tap one to travel.