Open for work & interesting problems

b43kd00r Rachit Verma

Senior Security Engineer · Pentester · Red Teamer

I break things on purpose so they do not break by accident. Founding security engineer at Anumana, securing AI that clinicians make decisions with and before that, five years of offensive work across government and consumer scale, including Red Team for the G20 Leaders Summit 2023.

  • 100+orgs recognised
  • 250+apps pentested
  • OSCP+eWPTXv2 · CAP
00

about

Who you are dealing with.

I am a security engineer with 5+ years in industry, and I have been hunting in VDP and bug bounty programs since 2017. The day job is product security for AI-driven healthcare. The instinct is still offence.

The work spans penetration testing, application and API security, cloud security, red teaming and increasingly LLM and agentic AI security. I have carried that across national infrastructure, consumer apps at India scale and regulated medical software. Three environments that fail in completely different ways, which is exactly why the combination is useful.

I like the findings nobody else bothers to chase: the ignored field, the parameter the server forgot to validate, the trust boundary that only exists in the architecture diagram. Most of my best bugs were logic flaws, not memory corruption.

  • 0critical vulns found
  • 0acknowledgements on record
  • 0faster incident detection
  • 0engineers mentored
01

experience

Where I have worked, and what actually shipped.

  1. Dec 2025 — Now current

    Founding Senior Cyber Security Engineer

    Anumana, Inc · AI-driven healthcare

    Regulated medical AI

    First security hire. I own product security for AI that clinicians make decisions with, where a bug is not a ticket, it is a patient-safety event. Led the end-to-end security assessment of WatchMate, the flagship medical AI platform backed by a Boston Scientific investment.

    Security-tested ChatECG and designed dynamic LLM guardrails driven by adversarial thinking rather than static pattern matching, then extended that into agentic AI reviews covering model misuse, data leakage, over-privileged tools and abuse of connected services.

    Built the secure SDLC from nothing: SAST/DAST, Semgrep and SonarQube code review, dependency scanning and continuous SBOM monitoring across every ECG AI pipeline. Stood up defence in depth with Sonatype Lifecycle and Nexus Firewall for artifact gating, Legit Security on the pipeline, Upwind for runtime, selected after a CNAPP bake-off I ran and Datadog SIEM for detection. Designed the GCP security architecture against HIPAA, ISO 27001 and FDA SaMD, threat modelled with STRIDE, DREAD and PASTA, hunted healthcare-specific threats, and run the bug bounty and VDP program.

  2. Jun 2025 — Dec 2025

    Security Engineer

    Curefit · House of Cult

    Consumer scale

    Cut average incident detection time by 30% by rebuilding alerting rules and thresholds around real attacker behaviour instead of vendor defaults and hunted APTs, insider threats and anomalous behaviour across the enterprise with Coralogix and custom tooling.

    Pentested web, mobile, network, Docker, Kubernetes, cloud and MCP surfaces, with 90% of critical and high findings remediated within 30 days. Automated CI/CD security with Gitleaks, TruffleHog, Bandit, dep-scan, Dependency-Check, Trivy and Docker Scout so secrets and container misconfigurations failed the build rather than the audit.

    Reviewed AWS and GCP against NIST and CIS benchmarks, researched and documented the Top 10 risks for LLM and generative AI applications to make AI adoption safe across engineering, ran the bug bounty and VDP program, and lifted overall security maturity by 30%.

  3. Aug 2021 — Jun 2025 ~4 years

    Security Engineer - Information Security & Audit Services

    C-DAC, Hyderabad · MeitY, Government of India

    National infrastructure

    Ran Red Team exercises for the G20 Leaders Summit 2023 and for PNB systems with no acceptable failure mode and no second attempt.

    Delivered 200+ web application and 50+ mobile application assessments for Fortune-scale Indian enterprises and government bodies across finance, energy, IT, defence and ministries, including SBI, IOC, ONGC, NPCI and BPCL. Led teams on engagement, presented findings directly to clients and owned remediation conversations through to close.

    Worked cloud (AWS, Azure, GCP), API security and Active Directory exploitation against OSSTMM, OWASP, NIST, PTES and CWE/SANS, mapped to NIST, CIS, ISO 27001 and RBI guidelines. Contributed to the Unified Blockchain Framework, the MeitY initiative behind the National Blockchain Service and mentored 9+ project associates through six months of VAPT and audit training.

  4. Jan 2021 — Jul 2021

    Security Analyst Intern

    C-DAC · Centre for Development of Advanced Computing

    Dark web research

    Assessed 80+ .onion sites for vulnerabilities using manual testing and automation, helped build a framework to identify and classify dark web content supporting forensic work for Indian law enforcement. Two published papers came out of this research.

  5. May 2026 — Now in progress

    Executive Post Graduate - Generative AI & Agentic AI

    IIT Kharagpur

    Education

    Going deep on the systems I am now paid to attack and defend. Preceded by a B.Tech in Computer Science from Chandigarh Engineering College, Mohali, and a Diploma in Computer Engineering from CCET Chandigarh. Bug bounty hunting started back then and never stopped.

02

skills

Rated honestly by how much I actually use them -> root is what I live in daily, user has shipped in real engagements, guest is working knowledge I can still be useful with.

Offensive

  • Web & API pentestingroot
  • Burp Suiteroot
  • Application securityroot
  • Nmap / Nessusroot
  • OWASP ZAP / SQLMaproot
  • Recon & OSINTroot
  • Red teaminguser
  • Active Directoryuser
  • BloodHound / PowerViewuser
  • Mimikatz / CrackMapExecuser
  • Ligolo-ng / Metasploituser
  • MobSF / Frida / Objectionuser
  • Hashcat / Aircrackuser
  • Reverse engineeringguest

AI & emerging

  • LLM securityroot
  • LLM guardrail designuser
  • Agentic AI securityuser
  • Prompt injection & misuseuser
  • MCP attack surfaceuser
  • OWASP LLM Top 10root
  • Blockchain / Web3guest

Defensive & automation

  • Secure code reviewroot
  • SAST / DASTroot
  • Semgrep / SonarQuberoot
  • DevSecOps pipelinesroot
  • Nucleiroot
  • Gitleaks / TruffleHogroot
  • Trivy / Grype / Syftuser
  • SBOM & supply chainroot
  • Sonatype Lifecycle / Nexususer
  • Threat huntinguser
  • SIEM — Datadog / Coralogixuser
  • Incident responseuser
  • Digital forensicsguest

Platforms & code

  • Linux / Bashroot
  • Pythonroot
  • GCProot
  • AWSuser
  • Dockeruser
  • Kubernetesuser
  • PowerShelluser
  • Azureguest
  • OCIguest

Governance

  • Threat modelling — STRIDEroot
  • DREAD / PASTAuser
  • Risk assessmentroot
  • ISO 27001root
  • Secure SDLCroot
  • NIST / CIS Controlsuser
  • IEC 62304 (SaMD)user
  • HIPAA / FDA premarketuser
  • RBI guidelinesuser
  • VDP / bug bounty opsroot
03

achievements

What actually changed because the work happened.

critical

Red Team - G20 Leaders Summit 2023

C-DAC · September 2023

Offensive assessment of systems supporting a summit of heads of state. Fixed date, no slippage and a threat model that included nation-state actors. Findings were remediated before the event opened. A separate red team engagement followed for PNB.

platform

WatchMate cleared for launch

Anumana · ongoing

Led end-to-end security assessment of the flagship medical AI platform backed by a Boston Scientific investment, full vulnerability testing through to remediation.

100+

Hall-of-fame acknowledgements

2017 — present · VDP & bug bounty

Vulnerabilities accepted and fixed by Microsoft, Apple, Sony, Intel, Zomato, IOTA, the Government of New Zealand, CERT-EU, CERT-IN and NCSC.nl, across web, mobile, API and cloud.

See the dated record
30% faster

Incident detection cut by 30%

Curefit · 2025

Rebuilt detection logic and tuned alerting around real attacker behaviour instead of vendor defaults. Less alert fatigue, earlier containment and 30% higher security maturity overall.

250+ apps

200+ web and 50+ mobile assessments

C-DAC · 2021 — 2025

Delivered across finance, energy, defence and ministries — SBI, IOC, ONGC, NPCI, BPCL-> while mentoring 9+ associates into working pentesters.

AI security

Guardrails for clinical AI

Anumana · ongoing

Designed adversarially-driven LLM guardrails for ChatECG and the agentic AI review process behind them, plus a GCP security architecture meeting HIPAA, ISO 27001 and FDA SaMD requirements.

Acknowledged by see all 49 with dates

MicrosoftAppleSonyIntelZomatoIOTANCSC.nlCERT-EUCERT-INGovt. of New ZealandNCIIPC+ 90 more
04

writeups

Written up the way I would hand them to you: the flaw, why it worked, what it was worth and the lesson a developer can actually use.

FND-001 critical

How ignoring one field led to Zomato account takeover

Zomato’s verification endpoint cared about the user_id and quietly ignored the code. One request with an empty code confirmed a stranger’s email and logged me in as them, no password, no token, no email access.

Program
Zomato
Class
Auth bypass · logic flaw
Disclosed
2018
Status
Fixed
Open the finding
FND-002 queued

More writeups in redaction

Several reports are still inside disclosure windows. They land here as they clear.

Ask me about them
05

certifications

Certifications and credentials, collected along the way.

offsec
OSCP+

OffSec Certified Professional

24-hour hands-on exam. Compromise the network or fail, no multiple choice.

OffSec · Dec 2024

web
eWPTXv2

Web App Pentester eXtreme

Advanced web exploitation and filter evasion, assessed by full report.

INE / eLearnSecurity · Jan 2023

appsec
CAP

Certified AppSec Practitioner

Secure design, code review and the OWASP risk landscape end to end.

The SecOps Group · Jan 2023

ai
IITKGP

Executive PG - Generative & Agentic AI

The systems I am now paid to attack, studied from the inside.

IIT Kharagpur · in progress

06

contact

Open for consulting, red team operations and collaboration.

What I take on

  • 01

    Application & API penetration test

    Web, mobile and API surfaces tested by hand, not just scanned. Authentication, authorisation and the business logic underneath, the flaws that tooling reports as a clean run.

    You get prioritised findings with reproduction steps, real business impact, a fix for each and a retest once they ship.

  • 02

    Cloud & infrastructure review

    AWS, GCP and Azure reviewed against CIS and NIST - IAM and privilege paths, network exposure, container and Kubernetes configuration, secrets handling.

    You get a mapped attack path through your own environment, ranked by what an attacker would reach first.

  • 03

    LLM & agentic AI security

    Prompt injection, model misuse, training and inference data leakage, over-privileged tools and the MCP surface. The work I do daily on clinical AI, applied to your stack.

    You get an adversarial assessment plus guardrail design that holds up to intent, not just keyword filters.

  • 04

    Red team & secure SDLC

    Goal-driven red team exercises against a real objective or building the pipeline that stops the findings recurring - SAST, DAST, dependency and secret scanning, SBOM monitoring.

    You get either an honest answer about what an attacker achieves or a build that fails before the audit does.

Scope, timeline and reporting format are agreed before anything starts. If you are not sure which of these you need, the first conversation is the one that sorts it out - and it is free.

You found the back door. Now let me find yours.

Hardly anyone scrolls this far. Since you did - penetration testing, application and cloud security, red team exercises, secure SDLC and LLM or agentic AI security. Not sure what you need yet? That is usually the best conversation to have.