Akshaya Lohia
  • UVA '30 Commerce + CS (intended)
  • FOUNDERS INC <3% Offseason acceptance
  • RUMA 150 waitlist users
  • DEMO DAY 1,500 attendees pitched
  • FOREST CREEK ~15 quants using my agents
  • TECHMUN ~1,000 students
  • BUDGET $40K+ run annually
  • HACKATHONS 2 wins, 2025 and 2026
  • MUN 8x Best Delegate

UVA '30, Commerce + CS, prev TJHSST

Akshaya Lohia

I'm a first-year at UVA, planning on Commerce and Computer Science. I'm very interested in both investing and applied AI. Currently helping small businesses implement AI in their work through Fiveafter.

Now

as of October 2026

Experience

  1. –

    RumaHome Search Platform

    FounderSan Francisco, CA

    • Founded and solo-built AI home-buying platform with Next.js, PostgreSQL, and Gemini; grew the waitlist to 150 users and launched beta
    • Integrated live MLS listings with AI scoring across safety, amenities, comps, and seller leverage to guide decisions for 3 beta homebuyers
    • Accepted to Founders Inc. Offseason (<3%); secured a live MLS data partner and pitched the platform to 1,500 attendees on demo day

  2. –

    Forest Creek Capital

    AI InternDallas, TX

    • Led AI development workshop for ~15 quants; designed pilot workflows that later expanded into broader enterprise-wide deployment
    • Designed AI agents used by ~15 quants to automate non-trading coding workflows, accelerate development and reduce repetitive manual work
    • Developed AI workflows to turn plain-English requests into automated energy/weather graphs and EPA/EIA infrastructure reports

  3. –

    Fiore Montessori School

    Technology InternVienna, VA

    • Designed 17-extension VoIP/intercom across 8 buildings and 2 campuses, saving >$7K annually while replacing vendor-quoted systems
    • Deployed 8-AP mesh Wi-Fi across 5 buildings, tripling coverage, sustaining 800 Mbps, and eliminating dead zones across both campuses
    • Installed 3 biometric/card readers across 2 campuses, enrolling 45 staff and consolidating access into one centralized keyless-access system

Projects

Autonomous AI Trading Fund System

A multi-agent system that researches, monitors a portfolio and plans trades overnight. Paper trading only. Active · paper trading Feb 2026 – Present
  • Built a multi-agent AI trading fund system automating overnight research, portfolio monitoring, investment decisions, and trade planning
  • Orchestrated macro, stock, idea, and fund-manager agents with senior review, role-specific prompts, and durable JSON/Markdown state
  • Integrated broker and market-data APIs to reconcile positions, compute risk, detect portfolio changes, and validate proposed trade plans

A systems-design project. No performance or return claims.

Agent architecture of my trading research system Macro and stock agents send their reports through a senior-review step. Reviewed reports and the idea agent's ideas reach a fund-manager agent, which keeps durable state. Its decisions go through risk checks that use broker and market-data APIs, producing a validated paper-trading plan. Macro agent Stock agent Idea agent Senior review Broker + data Durable state Fund manager Risk checks Trade plan Agent architecture of my trading research system Macro and stock agents send their reports through a senior-review step. Reviewed reports and the idea agent's ideas reach a fund-manager agent, which keeps durable state. Its decisions go through risk checks that use broker and market-data APIs, producing a validated paper-trading plan. Macro agent Stock agent Idea agent Senior review Broker + data Durable state Fund manager Risk checks Trade plan
Stock and macro research has to pass senior review before the fund manager acts on it.
What each part does
  1. Macro agent Writes the daily read on the macro backdrop.
  2. Stock agent Researches one company at a time.
  3. Idea agent Sources new ideas and keeps them in a queue.
  4. Senior review Grades stock and macro reports. Only A-grade work goes through.
  5. Broker + data Broker and market-data APIs: live positions, orders, prices.
  6. Durable state JSON and Markdown memory between runs.
  7. Fund manager Assigns the research, weighs it against the portfolio, decides.
  8. Risk checks Reconciles positions, computes risk, validates the plan.
  9. Trade plan Output: a validated plan. Paper trading only.
Role
Designer and builder
Stack
Python, OpenCode, Gemini, APIs
Status
Active · paper trading
Period
Feb 2026 – Present

Sidekick

Native macOS AI workspace grounding LLMs in personal files, Gmail and Calendar, local by default. Oct 2025 – May 2026
  • Built a native macOS AI workspace grounding LLMs in personal files, Gmail, and Calendar while keeping sensitive data local by default
  • Engineered hybrid retrieval with vector embeddings, BM25, Reciprocal Rank Fusion, and Qdrant for source-cited answers across local files
  • Orchestrated multi-step LLM agents with tool calling, lazy-loaded skills, output compression, and local/cloud model routing

Built on John Bean's open-source Sidekick. The original app frontend and some of its techniques come from his project. Sidekick by John Bean

Role
Designer and builder
Stack
Swift, SwiftUI, Qdrant, Ollama
Period
Oct 2025 – May 2026

Ruma

Founded and solo-built an AI home-buying platform. Beta launched · Founders Inc. Offseason May 2025 – Jan 2026
Ruma landing page
Ruma search with live MLS listings
Ruma buyer preferences form

Details under Experience

Role
Founder, solo engineer
Stack
Next.js, PostgreSQL, Gemini, MLS data
Status
Beta launched · Founders Inc. Offseason
Period
May 2025 – Jan 2026

Aura

A programmable acoustic shield that detects and deters wildlife from crops. I built the iOS app in Swift and its wireless link to the Arduino firmware. 1st place, Hack The Nest 2025 Sep 2025
Exploded view of the Aura acoustic deterrent
Aura device render
The first Aura prototype, top view
Role
iOS app and wireless link (team of 3)
Stack
Swift, iOS, Arduino, mesh networking
Status
1st place, Hack The Nest 2025
Period
Sep 2025

MyHouse

A tool that helps real estate investors find investment properties matching their criteria. Best Finance Hack at HackTJ 2026. Best Finance Hack, HackTJ 2026 2026
Role
Team of 3
Status
Best Finance Hack, HackTJ 2026
Period
2026

Vitrio

iOS app that turns a photo of an everyday object into an explorable 3D technical model, with internals you can isolate, explode and inspect. Prototype Jul 2026
Role
Designer and builder
Stack
Swift, SwiftUI, iOS
Status
Prototype
Period
Jul 2026

Bastion

iOS assistant that indexes Gmail and Google Calendar with on-device embeddings for semantic search and a daily briefing. Prototype Jun 2025
Role
Designer and builder
Stack
Swift, SwiftUI, Core ML, Google APIs
Status
Prototype
Period
Jun 2025

Albert

A personal-finance app built at HackTJ 2025. HackTJ 2025 build Mar 2025
Status
HackTJ 2025 build
Period
Mar 2025

Blitz

A real-time multiplayer word game. May 2025
Period
May 2025

EcoStruv

Struvite-synthesis process for treating agricultural wastewater and recovering nutrients; MIT URTC poster. Poster, MIT URTC 2023 Nov 2022 – Oct 2023
EcoStruv treatment unit render
EcoStruv unit render, outdoors
  • Led 4-person team developing EcoStruv, a struvite-synthesis process for treating agricultural wastewater and recovering nutrients
  • Analyzed product economics and commercialization strategy, translating working demos into a business case for farm treatment
  • Selected for poster presentation at MIT URTC, an IEEE undergraduate research conference hosted at MIT alongside university teams
Role
Team lead, 4-person team
Status
Poster, MIT URTC 2023
Period
Nov 2022 – Oct 2023

Leadership

  1. Sep 2026 – PresentCharlottesville, VA

    Alternative Investment Fund at UVA

    Analyst

    Student-run investment fund at UVA (founded 2007) that researches and pitches trades across its Fundamental Equities, Global Macro and Special Situations funds.

    • Completing AIF's full analyst training program: learning to understand a business and research and manage public investments
    • Rotating through the Fundamental Equities, Global Macro and Special Situations funds; writing company write-ups

  2. Sep 2026 – PresentCharlottesville, VA

    Virginia Consulting Group

    Analyst

    Student-run consulting group at UVA whose analysts work in semester-long teams on pro bono and paid client projects.

    • Consulting for Midnight Ventures, a wellness-focused VC firm, on a semester-long project team; final client presentation in December
    • Completing VCG's consultant analyst program, including weekly new-analyst training in research and communication fundamentals

  3. May 2022 – May 2026Alexandria, VA

    TJ Model United Nations

    Secretary-General (President)

    • Ran TechMUN, the East Coast's largest high school Model UN conference, coordinating operations for approximately 1,000 students
    • Managed a $40,000+ annual budget, coordinated 80+ staff, and oversaw year-round operations for a 150-student Model UN program
    • Conducted Model UN training for 100+ delegates, strengthening conference skills across a nationally recognized program

  4. May 2024 – May 2026Alexandria, VA

    TJ Emerging Technologies Club (Commas Club)

    President

    • Led a 30+ member high school club meeting weekly to teach students practical uses of emerging technologies and AI tools

  5. Nov 2022 – Oct 2023Alexandria, VA

    MIT Undergraduate Research Technology Conference (URTC)EcoStruv

    Team lead (4-person team)

Contact

Email is the fastest way to reach me. I'm always happy to talk about investing, AI, or whatever you're building.