Manohar Papasani
AI R&D · Hyderabad

Manohar Papasani

GenAI Lead & AI Architect at Minfy · AI Systems Builder

I design and build AI systems that do real work.

I lead a team at Minfy building agentic GenAI solutions on AWS. In my R&D lab, a digital workforce of 35 AI agents researches, writes, reviews and publishes around the clock, and I publish open research on making agents reliable. Before Minfy, I spent five years at Bank of America leading GenAI and automation for a delivery organisation of 100 to 150 people.

See the work Get in touch
Agentic AI on AWSEx-Bank of America AVPIIT Patna certified, 2026
IIT Patna certificate: Generative AI and Agentic AI for Developers, awarded to Manohar Papasani, October 2026 Certified by IIT PatnaGenerative & Agentic AI for Developers · 6 months · 2026 Claude Certified Architect · preparingOutskill GenAI Sprint
Bank of America28Awards in five years, for GenAI leadership and delivery1 Platinum8 Gold7 Silver12 Bronze
Bank of America14,300Engineer-hours saved every year with GenAI automation
In progress · goal 20035 / 200Digital employees on duty, building toward 200
Certified · 2026IIT PatnaGenerative & Agentic AI for Developers, six months
Reel 1 · The workforce

Nobody here takes a day off

Every agent below runs on its own schedule in the cloud, with a quality gate before anything goes out and a trace of every step. The goal is 100 of them working together by 31 January 2027, and 200 within a year.

35 on duty todayGoal 100 by 31 Jan 2027 · 200 by Oct 2027
Reel 2 · Architecture

How the digital workforce runs

The systems behind the work: what triggers each shift, which models and tools the agents use, where quality is checked, and how every step is traced.

01 · SYSTEM MAP

Seven silos, and no row calls another

Each team has its own exact-time trigger on Cloudflare, its own workflow on GitHub Actions and its own agents. If one team breaks, the other six do not notice.

When (IST) Clock · Cloudflare Worker Run · GitHub Actions Agents · LangGraph Output 8:00 AMLeadership ace-leadership-trigger motivation-daily.yml Leadership team 12:00 PMAI interview Q&A ace-interview-trigger interview.yml Interview team 4:00 PMAI news digest lbe-trigger daily.yml News pipeline + Haiku gate 11:30 PMTech learning ace-learning-trigger motivation.yml Learning team LinkedIn 4 posts a day 5:35 AMMon and Thu ace-brand-analytics-… brand-analytics.yml Brand analytics agents Website metrics 9:00 AMOperations ace-operations-trigger cost-report · anomalyweekly-evals (Sun) Cost, anomaly, eval agents Reports and alerts +10 to +90 minAfter every slot promo-trigger distribute.yml Promotion engine X Bluesky Facebook Dev.to same post, read-only, staggered Shared by no one: each row has its own Worker, timer, workflow, self-heal and alerts. A GitHub watchdog checks all of them on a separate clock. No post is scheduled by GitHub cron: Cloudflare fires on the minute, GitHub only runs the work.
Seven pipelines, one per row. The amber lines are the four daily posts; the blue line is cross-posting, the only link between rows, and it only reads post records.
02 · INSIDE A POST TEAM

Every post is written, checked and either published or dropped

The leadership team as it runs in code: a LangGraph state graph. Routing is decided in code, not by the model, so a weak post is skipped instead of published.

quote not verified: find another fails review: one rewrite, at most Briefinsights Plannerpicks the theme Quote scoutfinds a quote Verifierchecks Wikiquote Writerowner's voice Poster scoutmakes the card Criticfour judges Publishexactly once pass Critic's judges Fact-checker Technical reviewer Consistency Generic-voice check Skip nothing strong enough, quote unverifiable, or second review failed: the slot heals or stays empty, never weak
Same writer, critic and publisher in every team. The interview team swaps the first three steps for an interview scout that finds a real reported question; the news team swaps them for a news scout.
03 · ONE DAY

Fixed slots, then a calm ripple to every platform

Each post goes to LinkedIn on the minute. The other platforms follow one at a time, at the same offsets every day, and never as a burst of old posts.

24 hours, IST 00 06 12 18 24 8:00 Leadership 12:00 Interview 16:00 AI news 23:30 Learning 5:35 analytics (Mon, Thu) 9:00 cost and anomaly checks 23:45 billing Every slot, zoomed in: minutes after the post LinkedIn+0 X+10 Bluesky+25 Facebook+50 Dev.to+90 At least 45 minutes between two posts on one platform · posts older than 12 hours are never sent
Four slots, five platforms, twenty posts a day. Every post carries the same six follow links: LinkedIn, GitHub, Bluesky, X, Dev.to and Facebook. On X the links are written as handles, because X posts carry no links.
04 · SELF-HEALING

A missed slot repairs itself within minutes

After each slot the team's own Worker checks every five minutes. If no run has succeeded and none is running, it starts another. Alerts fire only if healing has not worked.

Minutes after the slot Trigger on the minute a check every 5 minutes from +5 +30 alert +60 alert +90 final alert Heal only when no run succeeded and none is running · at most 4 healing runs · posted exactly once, recorded in a ledger GitHub watchdog checks every pipeline on GitHub's own clock, so a Cloudflare problem cannot hide itself
Measured on 3 November: the 12 PM post fired at 12:00:54 and the 4 PM post at 4:01, and every platform followed at its exact offset.
05 · QUALITY LOOP

Every agent is tested on its own, every week

Each judge runs with its production prompt on planted mistakes and on clean answers. It must catch real faults without crying wolf, or the owner is alerted.

Per-agent testsSundays, 2:00 AMplanted faults + clean cases Score each agentcatches ≥ 75% of faultsfalse alarms ≤ 34% Fix on a branchprompt, schema or modelfallback CI gatelint, types, 228 tests,scheduler tests Merge, then tag stablebackup tag before every change,one-click rollback next Sunday, the fixed agent is measured again First finding, 3 November: the interview reviewer flagged both clean answers. After the fix: no false alarms in three runs, both planted faults still caught.
Measured, not claimed. A judge that crashes counts as a false alarm, so a broken agent can never pass its test.

Open the diagrams on their own page

Reel 3 · Lab notebook

What we found this month

Dated entries from the lab. Results marked preliminary are still being run or written up.

    Reel 4 · The arsenal

    50 repositories. One builder.

    Research, production agent systems, free courses anyone can take, and open-source contributions. Public ones link to the code; private ones run real workloads and stay closed.

    50Repositories on GitHub
    31Public, open to anyone
    0Free courses for everyone
    0Systems running live
    Credentials

    Certified to build what I build

    Formal training behind the hands-on work. The newest: a six-month programme in Generative and Agentic AI from IIT Patna, completed in October 2026.

    Completed · Oct 2026 Certificate from the Indian Institute of Technology Patna: Manohar Papasani has completed the six-month certificate course Generative AI and Agentic AI for Developers, issued 29 October 2026.
    • Completed · 2026Generative AI & Agentic AI for DevelopersIndian Institute of Technology Patna · Continuing Education Programme · 6 months, online
    • Continuous learning · Jan 2026Generative AI Sprint: AI for Founders and BusinessOutskill · 14 days · AI Generalist Accelerator Program (NSDC course: AI for Founders and Business)Prompt Engineeringn8n AutomationNo-Code ProductsVoice AgentsMCPMultimodal AIView certificate
    • Preparing · Dec 2026Claude Certified Architect – ProfessionalAnthropic · Hands-on labs under way
    • CertificationCisco Certified Python ProfessionalCisco
    • PostgraduateMaster of Computer Applications (MCA)VRIPS · Sri Venkateswara University

    Certificate and enrollment numbers are hidden for privacy; available on request.

    Reel 5 · The journey

    From a farming village to a digital workforce

    I build software that does real work. For more than ten years that meant automation, data platforms and production systems. Today it means AI agents: digital employees that research, write, check and publish on their own.

    1. Roots

      A farming village

      I grew up in Komarapudi, near Nellore, in a farming family. Education was the way forward, all the way to an MCA from Sri Venkateswara University.

    2. 2010 – 2019

      The builder years

      In Hyderabad I built Python automation at MoSys, Java platforms at Pyramid Softsol, and Spark and Scala pipelines on AWS at Virtusa.

    3. 2021 – 2026

      Automation at scale

      Five years at Bank of America, first in SRE and DevOps, then leading GenAI for a 100–150 person organisation, where I founded the prompt-engineering team.

    4. 2026 – now

      A digital workforce

      At ValueMomentum I built the agent factory behind a 20-agent programme. Today I lead agentic GenAI on AWS at Minfy and run this lab alongside.

    The question changed from "can AI help an engineer?" to "can AI do the job, reliably, every day?"

    Teaching has run alongside it all. I have been a trainer since 2004 and was an Assistant Professor. I have trained more than 500 engineers and run a learning group in every company I join; the free courses here come from the same habit.

    Experience

    1. Oct 2026 – presentHyderabad

      Minfy Now

      AI Developer – Lead

      • Lead a team of four engineers building agentic GenAI solutions on AWS Bedrock for AWS Marketplace.
      • Own the build: environments, scope, architecture and delivery of the first solutions.
    2. Aug 2026 – Oct 2026

      ValueMomentum

      GenAI Solution Architect · AI Senior Lead

      • Designed the agent factory behind a client's 20-agent incident-management programme: a 2-year plan delivered in 6–8 months, with ~70–80% platform reuse.
      • Built the flagship root-cause-analysis agent on Copilot Studio: time to first answer from ~30 s to under 5 s, 1,506+ incidents per run, zero new infrastructure cost.
      • Brought CI, quality gates and a regression harness to low-code agents; continuous evaluation lifted answer consistency from ~25% to ~80%.
    3. Apr 2021 – Jul 2026

      Bank of America

      AVP · GenAI & Automation Lead · SRE/DevOps Lead

      • ~14,300 engineer-hours a year saved: a prompt-engineering framework that automated 28,600+ compliance reviews a year at 4× throughput (40 → 10 minutes each).
      • Founded the prompt-engineering team and built a library of 10,000+ structured prompts; 100+ automations (20+ built, 80+ mentored across 8–10 teams).
      • Database health checks across 27 databases cut from 1.5 hours to 5–10 minutes; incident management and service restoration in critical windows.
      • Led Copilot adoption for a 100–150 person organisation; GenAI orchestration (LLMOps) with LangSmith and Galileo.
      • 28 awards in five years, including Platinum for GenAI leadership.
    4. Aug 2019 – Mar 2021

      Virtusa

      Senior Consultant

      • Spark and Scala data pipelines on AWS EMR for a research-analytics company: automated Hive ingestion, large-scale S3-to-HDFS migration.
      • Spark applications with JanusGraph and HBase for an insurer's disability-evaluation platform.
    5. Dec 2012 – Sep 2014

      Pyramid Softsol

      Software Engineer

      • Java, Spring and Hibernate application layers for two client platforms; automated data migration; production incident fixes.
    6. Oct 2010 – Jul 2011Hyderabad

      MoSys India

      Technical Consultant

      • Python automation for semiconductor test data (CSV, JSON, XML); Oracle extraction, validation and cleansing.

    Teaching

    1. Aug 2011 – Nov 2012Nov 2014 – Jul 2019

      Independent technical trainer

      Python, Java and big data · Corporate and campus · 500+ engineers trained

    2. Nov 2007 – Mar 2010

      DRK College of Engineering and Technology

      Assistant Professor

    3. 2004 – 2007Nellore

      CAT Computers

      Trainer · Where teaching started

    Education

    1. Postgraduate

      MCA

      VRIPS · Sri Venkateswara University

    2. Undergraduate

      Bachelor's degree

      Vivekananda Degree College · Sri Venkateswara University

    3. Intermediate

      Ratna Junior College

    4. Classes 8 – 10

      Vidya Bharathi

    5. Classes 1 – 7

      Siddhartha English Medium School

    10+Years in software, data platforms and AI
    14,300Engineer-hours a year saved at Bank of America
    28Awards, including Platinum for GenAI leadership
    20Agent programme delivered in 6–8 months instead of 2 years

    MCA · Cisco Certified Python Professional · Generative AI & Agentic AI for Developers, IIT Patna (6-month certificate, 2026) · Claude Certified Architect – Professional (preparing)

    Coming next

    Back the lab

    Looking for partners, sponsors and early users for the next builds. Each one ships in public with measured results.

    Talk to me on LinkedIn See the code
    • Agent swarm · A factory that turns a short spec into a working agent, and a shared board where 100 agents coordinate
    • Guardrails agent · Interviews a team, then enforces their rules across every agent
    • Competition autopilot · Finds AI competitions worldwide, builds entries, fixes its own failures
    • ML inside agents · Routers, rankers and guards that make agents cheaper and safer