Aman Bhandula

Some things about me:

• Founder, operator, and AI systems builder at Farmako AI
• I build agent-native tools, research systems, neurotech experiments, and practical operating software. This site now tracks the open-source portfolio I am preparing from recent work.
• Building from India
• Open to product, infra, and agent collaborations
• Hi, I am Aman Bhandula. I work on applied AI systems that can own real workflows: agents, local skills, research loops, internal tools, creative automation, and healthcare operations.
• The current focus of this site is a private-first open-source portfolio. Each project below has been recovered from local or iMac source, cleaned into its own repository, documented, and tested before release review.
• I care about tools that are ambitious but operationally honest: clear setup, no leaked credentials, reproducible checks, and claims that match what the code actually does.
• Farmako AI
• OpenClaw
• Product Systems
• AI Operations
• Reach me at aman@farmako.ai

Some things I am building:

• NeuroState Lab
  • Screen-aware neurotech desktop app with observation mode and calibration mode for EEG + fMRI-style dataset creation.
  • Aman Bhandula. Neurotechnology systems project, 2026
  • Electron app + FastAPI backend + TRIBE v2 modeled cortical responses + exportable calibration dataset manifests
  • Calibration mode now time-locks controlled task blocks, manual labels, screen/audio chunks, TRIBE fMRI-like arrays, and EEG samples into an exportable research manifest. TRIBE outputs remain modeled population-level estimates unless replaced with validated measured fMRI hardware.
  • Links: Private repo, README, Calibration docs, Calibration screenshot
• Images to 3D Viewer
  • A Codex skill and browser viewer for turning image sets into inspectable 360/panorama scenes.
  • Aman Bhandula. 3D viewer tooling, 2026
  • Includes a verified WebGL example and deterministic local viewer generator
  • Built from scratch as a reusable skill with generation, local serving, canvas-pixel checks, and interaction smoke tests.
  • Links: Private repo, README
• Codex Google Ads Manager
  • A reusable Codex skill and API wrapper for safer Google Ads inspection and campaign operations.
  • Aman Bhandula. Marketing operations automation, 2026
  • Read-only health checks, dry-run guardrails, setup docs, and credential-safe defaults
  • Generalized from live operations work; no account secrets or Farmako account identifiers are included.
  • Links: Private repo, README
• Codex Dreaming Agent
  • A generic local-learning loop that turns recent Codex work into compact operating context.
  • Aman Bhandula. Agent memory and workflow project, 2026
  • Daily reports, personalization updates, skill queueing, and validation gates
  • Packaged as a reusable Codex skill rather than an Aman-specific private workflow.
  • Links: Private repo, README
• Autoresearch Advanced
  • An advanced autonomous research loop extending Karpathy-style autoresearch with RALPH and deep-research review.
  • Aman Bhandula. Autonomous research systems, 2026
  • Plan, dataset, cleaning, experiment, and review agents working in a continuous loop
  • The RALPH layer adds periodic analysis so the system improves research direction instead of only running random experiments.
  • Links: Private repo, README
• AgentArxiv
  • An agent-first research publishing platform for autonomous agents to publish ideas, experiments, and reviews.
  • Aman Bhandula. Agent research infrastructure, 2026
  • Recovered from the iMac source tree and packaged as a sanitized private repo
  • Includes API-key agent access, papers, projects, milestones, bounties, comments, reviews, run specs, and SDK/API surfaces.
  • Links: Private repo, README
• Farmako Animation Generator
  • A branded animation and illustration generator for Farmako design systems and motion assets.
  • Aman Bhandula. Creative automation tool, 2026
  • Next.js studio with generation, local library, video workflow, and skill packaging
  • Published as a private personal repo for review first; environment variables and generated media are excluded.
  • Links: Private repo, README
• OpenClaw 100x Guide
  • A practical operating guide for setting up an OpenClaw machine, workflow, accounts, and daily agent cadence.
  • Aman Bhandula. OpenClaw operating manual, 2026
  • Machine setup, model choice, separate accounts, Kanban workflow, and daily calibration habit
  • Guide-only repo with templates and verification checklists; it does not mutate the upstream OpenClaw worktree.
  • Links: Private repo, README
• Open-source portfolio packaged into private GitHub repos.
  • Recovered, sanitized, tested, and pushed eight recent projects into separate private repositories under Amanbhandula for review before public release.
  • 15-05-2026

Some things I am thinking about:

• How do we make agents reliable enough to own real workflows across browser, desktop, API, code, and deployment surfaces?
• How can autonomous research systems plan better experiments instead of only running more experiments?
• How should neurotechnology interfaces communicate modeled brain-state estimates without overstating what is actually measured?
• People I enjoy building with
  • Collaborators, interns, and operators who have shaped my thinking.
  • Design collaborators: Interface systems, brand direction, product polish
  • Engineering partners: Infrastructure, tooling, full-stack execution
  • Operator peers: Workflows, GTM systems, internal process design
• Startups and teams I track closely
  • Vertical software, AI ops, healthcare tooling, and systems software.
  • Farmako AI: Healthcare and operations systems
  • Agent tooling ecosystems: Skills, runtimes, orchestration, and infra
• People who have sharpened my taste
  • Builders who push for clarity, quality, and leverage.
  • Founders who sweat detail: Product quality, shipping discipline, and sharp communication
  • Operators who simplify complexity: Execution systems, process design, and calm decision-making
• Themes I keep returning to
  • The work I want to keep doing over the next decade.
  • Agent-native products: Interfaces and systems where automation can own real work
  • Workflow infrastructure: Permissions, review, monitoring, and operational rigor
  • Vertical AI software: Products with strong outcomes in high-friction domains

Some things I have written:

• fMRI for BCI Task Design: From Voxels to BOLD Models
  • A visual tutorial connecting voxels, BOLD signals, task blocks, and multimodal BCI dataset design.
  • May 26, 2026
• OpenClaw 100x Guide
  • Aman Bhandula
  • Operating guide, 2026
  • Machine setup, model selection, separate accounts, Kanban, and a daily conversation cadence for a serious OpenClaw workflow.
• RALPH and deep-research loops for Autoresearch Advanced
  • Aman Bhandula
  • Research systems note, 2026
  • A continuous research-agent design that adds periodic deep analysis to planning, dataset creation, cleaning, and experiment execution.
• NeuroState Lab safety and interpretation notes
  • Aman Bhandula
  • Neurotechnology project note, 2026
  • Documentation for observation mode, calibration mode, EEG/fMRI integrations, and honest interpretation of TRIBE-modeled cortical responses.

Some links:

• Email
• GitHub
• LinkedIn
• X / Twitter