Research profile

Shreyash Gondane

Independent researcher · Software engineer, New York

I build and study AI agents for financial research that can be checked: every number traced to its source, only data that existed at the time of a call, and every call recorded before its outcome is known.

Research statement

LLM agents are increasingly asked to read markets and explain their calls, but their outputs are hard to audit: numbers appear in prose with no source, backtests quietly use data that did not exist yet, and track records can be edited after the fact.

My work treats verifiability as the core design constraint. I turn every number an agent reports into a structured claim that points at the exact value in a stored, point-in-time source snapshot, check each claim before anything is published, and record every call in an append-only, hash-chained ledger so the record can be scored forward only.

I also study how to build such systems cheaply and reliably with agents themselves: a builder/critic loop in which low-cost models write code and a stronger model plans and reviews, recording every attempt as labelled data for distillation.

Claim-level provenance

Every reported number is a claim with a JSON pointer into a hashed source snapshot, verified before publication.

Point-in-time evaluation

Agents only see data fetched before their decision time; calls are scored only after they are recorded. No look-ahead leakage.

Tamper-evident records

Append-only, hash-chained call ledgers with database-level guards, so a track record cannot be edited after the outcome.

Distilling coding agents

Builder/critic twin loops that produce labelled trajectories, corrections and failure modes for training smaller models.

Writing

Receipts for every number: claim-level provenance and point-in-time verification for LLM trading agents

in preparation

Method and early results from GlassBox: the baseline discrepancy audit, the verification gate, and the live forward-only record.

Projects

GlassBox

2026 – presentactiveLLM agentsverificationquant financeopen source (MIT)

An evidence-first multi-agent investment-research committee. Release s3-part1 adds a point-in-time snapshot store for every data fetch, structured claims with exact source pointers, a verification gate that re-checks every number (source match, point in time, staleness, snapshot integrity, units), a compliance filter, a gate-health dashboard and an append-only hash-chained ledger. Research preview only; not investment advice.

Twin-agent build lab

2026 – presentactiveagent orchestrationdistillationdatasets

Building production software with a free model as builder and a strong model as planner and reviewer. Every run is traced; accepted runs, rejections and before/after correction pairs form a labelled dataset. Includes CPU benchmarks of small local coding models (Qwen2.5-Coder 1.5B at ~42 tokens/s on a 4-core laptop CPU).

Multi-agent trading system

2026earliermulti-agentvalue function

An earlier multi-agent investment system built around a V(n) value function, with dashboards and analytics. The groundwork that GlassBox's verification-first design grew out of.

Scientific discovery agent

2025earlieragentsresearch automation

Infrastructure for an agent that assists scientific discovery workflows.

Track record

A dated log of releases, results and service, each linked to evidence where it exists. Newest first. Follow via feed.

  1. Release

    GlassBox: gate-health dashboard, stricter verification gate, compliance filter

    Snapshot-integrity and unit checks added to the verification gate; an A7 compliance filter with 113 rule tests; a dashboard of the daily verification pass rate.

  2. Milestone

    First verification-gate results on live production data

    Every number in the first live committee decisions re-checked against its source snapshot; decisions provably unchanged by the verification layer.

  3. Release

    GlassBox s3-part1: verification foundations

    Point-in-time snapshot store, structured claims, A6 verification gate, hash-chained call ledger, feature flags and kill switches.

  4. Benchmark

    Local coding-model benchmark on a laptop CPU

    Qwen2.5-Coder 0.5B and 1.5B (Q4_K_M, llama.cpp): ~68 and ~42 tokens/s generation on an i7-1165G7.

  5. Open source

    GlassBox made public under the MIT license

    Repository opened with a README comparing it honestly with TradingAgents.

Service

Reviewing and talks will appear here.