I'm Shreenivas Sangoram. I design agentic systems, RAG pipelines and AI-assisted investigations for firms where a wrong answer is expensive, and I open up small language models to see why they behave the way they do.
An agent reads a model's tensors, forms a hypothesis and patches safetensors directly. Tensor-Slayer made 44 targeted edits to Qwen-0.6B.
Tensor-Slayer ↗Sycophancy is a straight line in a small model's hidden states. Subtracting it took Qwen3-0.6B from 12.5% to 53.8% on pressure prompts.
Read the write-up →RL and distillation on 0.6B–4B models. Claude diagnosed a 0.8B model's tool-call failures and wrote LoRA fixes: 42% → 75% strict accuracy.
See the lineage →An agent drives an iPhone over USB from Linux, no Mac needed. A signed Android client keeps a running agent in your pocket.
Piphone ↗The Codex desktop app on Linux (57★), a kernel that schedules agent workers, and a terminal that draws the session as a tree of light.
Codex-App-Linux ↗Generative media where the model writes an editable score and a deterministic engine renders sound, or a million points of light.
See the work →
About
I came to AI from the risk side. I spent years at Deloitte and PwC on fraud investigations, AML, integrity due diligence and compliance reviews, and ran supplier risk across a $35M commodity book at Whirlpool.
Since 2022 I've run ConsultingAI, building the AI that this kind of work needs: systems that cite their sources, flag what they don't know, hold up in an audit, and get a C-suite to say yes.
Outside client work I research model behaviour: steering activations, patching weights directly, building benchmarks. I also build agent tooling for phones, desktops and terminals, and render films of light with raw CUDA.
Research agents, multi-agent orchestration and retrieval pipelines, designed to run securely on AWS Bedrock/SageMaker or on-prem.
AI-assisted fraud investigation, integrity and senior-management due diligence, and market intelligence. Findings come out structured, sourced and defensible.
Audits of enterprise AI deployments, model-behaviour evaluation, prompt-injection testing and governance aligned to GDPR/CCPA, plus custom benchmarks.
A mix of interpretability research, developer tooling people actually install, and engineering done for the joy of it.
Is sycophancy a straight line in a model's hidden states? Linear probes say yes, and the signal peaks at 54–64% of network depth. Subtracting that direction during generation reduces sycophantic agreement with no RLHF and no fine-tuning.
An agent reads a model's static weights, forms hypotheses and patches safetensors directly, which amounts to interpretability without inference. Case study: 44 targeted tensor edits to Qwen-0.6B, evaluated on MMLU and HumanEval.
A benchmark for small decision models that return typed answers with a probability for every option. It scores four things in one number: chance-corrected intelligence, calibration, speed and cost.
A one-script installer that turns the official macOS Codex desktop DMG into a working Linux app. It rebuilds native modules, stubs the macOS-only ones, and gets mobile pairing and in-app Browser Use running.
A signed Android client for a running Hermes Agent dashboard, connected over Tailscale with a single npx command. It keeps your agent in your pocket without reimplementing it.
Give your iPhone a task in plain language and an agent carries it out over USB through WebDriverAgent, with guarded posting and long-running sessions. It works from Linux, no Mac needed.
Jeff Dean and Sanjay Ghemawat's Performance Hints, packaged as a Claude Code plugin that gives coding agents guidance on profiling, code review, testing and API design.
A deterministic film studio written in raw CUDA (nvcc + ctypes, no PyTorch at render time). It turns a picture into a million points of light, and a score plus assets render to a finished reel in one pass.
A native terminal rendered as a 3D object, with the coding-agent session drawn as a tree of light: silver for the past, gold for the live tip. You can sail through the session and ride back along its paths.
Sound composed as SVG scores that an LLM writes and edits. The score is the program and the audio is rendered from it, which led to a non-neural voice-conversion experiment.
Fifty standalone web-app tasks under one shared contract (two viewport widths, keyboard-only use, accessible errors), and a Playwright verifier that grades through the rendered UI and separates binding, infrastructure and behaviour failures.
Pocket Pi runs the Pi agent loop inside iOS WebKit, with Swift bridges for the network and the phone's own tools.
Agent-0, a Claude Code plugin that runs as a kernel with process state, syscalls and a scheduler for Claude, Codex and OpenCode workers, and a neural bus where agents in terminal panes message each other and stalled replies get nudged.
Two years of putting agents on Android: an LLM agent driving apps through Appium, an agent TUI on mobile-mcp, an embedded agent runtime prototype, and a signed Android client for a running Hermes agent.
A C runtime for Gemma 4 12B QAT on an RTX 3050 Laptop GPU, layer streaming from disk at exact safetensors offsets, and a 27B project held to a strict correctness contract.
Standing up Datacurve's DeepSWE benchmark (113 long-horizon tasks in the Harbor format, sandboxed with Pier) for MiniMax M3, with my own harness configs and a CLI harness adapter.
An implementation of ARPO (Agentic Reinforced Policy Optimization) for Qwen3-0.6B: SFT then RL, entropy-based adaptive rollouts, step-level credit for tool calls and the Muon optimizer, plus the numerical-stability fixes RL training needed.
A training codebase that aligns a 0.8B model with GRPO, rewarded by an LLM judge that returns typed decisions instead of free-text grades. Implementation complete and tested; no training run yet.
A data pipeline, harness contract, probes and evaluation suite for teaching Qwen3-0.6B to operate a coding agent, built on exact tokenizer measurements and deterministic structural gates rather than guesses.
Notes from working OpenAI's Parameter Golf challenge with coding agents as the research staff: a Modal lab, proxy runs, and labs for tensor interventions, RL and mixture-of-experts ideas.
A safety evaluation where two AI agents get what they believe is a private, encrypted channel and are prompted to discuss kill switches; the harness reads the conversation anyway and scores agreement to resist or evade shutdown.
Market-simulation environments for training and evaluating trading agents, built by someone who has written an LLM trading bot runtime and RL traders.
Describe a video and an LLM writes a score (a timeline of events) and the renderer that plays it; the studio renders the video with FFmpeg and checks that everything on screen came from the score.
Each post opens with a short answer, then shows the evidence. All writing →
Every folder on my drives, catalogued and dated by git history and agent transcripts. Most are experiments; a few became products. This is the shape of the work, followed by how each line of it evolved.
Projects started per month · peak
Agentic RAG, research agents, multi-agent orchestration, enterprise AI audits, AI-assisted fraud investigation, IDD/SMDD and market intelligence. I architect the platform, present the strategy to the C-suite, and hire and lead distributed engineering teams.
Fraud investigations, AML, compliance reviews and risk consulting across e-commerce, aviation, pharma, energy, NBFC, infrastructure and more.
Owned Sourcing Excellence: supplier compliance under FCPA and the UK Bribery Act, supplier risk management, and $35M of injection-moulding spend.
Integrity due diligence ahead of partnerships, M&A and strategic decisions across a dozen sectors.
Quality on engine castings and T6 heat treatment, and customer representative to Tata Motors.
MBA, Competitive Intelligence (Amity) · B.Tech (BVBCET Hubli)
PyTorch · Transformers · TRL / GRPO · LoRA & PEFT · TransformerLens · ROME / MEMIT · CUDA
Claude · LangGraph · CrewAI · AutoGen · smolagents · Google ADK · MCP · A2A
AWS Bedrock · SageMaker · Lambda · Kinesis · Step Functions · Docker · Kubernetes · Azure
Python · Rust · Go · C · SQL · Spark · Pinecone · Milvus · Neo4j · Snowflake
Whether you're scoping an AI build, auditing one that's live, or need an investigation done faster and to a higher standard, I'd like to hear about it.