SKM/PORTFOLIO
Product Analyst (in practice)Systems ThinkerTechnical Leader

I turn operational complexity into product clarity.

I do Product Analyst work from inside enterprise support — and I have the shipped platform to prove it. I investigate complex product problems, run campaigns and A/B tests end-to-end, uncover hidden patterns in data, and turn recurring operational issues into scalable product improvements. Most recently, I architected and built an AI ticketing platform solo that cut triage time from 30 minutes to 30 seconds.

View product case studies →Let’s build better products
Tech Support Lead → Product · Bengaluru
Suraj Kumar Mohapatra
SURAJ KUMAR MOHAPATRA·TECH SUPPORT LEAD → PRODUCT·BENGALURU
01 / Capabilities

What I bring to a product team.

I haven’t just supported a martech platform — I’ve operated it: building segments, configuring automations, launching campaigns, and A/B testing them across 15+ enterprise accounts.

DEPTH BY DISCIPLINE
Root cause & product discoveryExpert
SQL & data investigationExpert
Enterprise SaaS / CDP platformsExpert
AI, RAG & LLM evaluationAdvanced
Dashboards & reportingAdvanced

Self-assessed working depth — how independently I operate in each area today.

Product thinking

  • Product discovery
  • Root cause analysis
  • Prioritization
  • Requirements gathering
  • Journey mapping
  • Stakeholder alignment

Data & analytics

  • SQL
  • Snowflake
  • MySQL
  • MongoDB
  • Power BI
  • Python
  • Advanced Excel

Enterprise SaaS

  • Customer Data Platforms
  • Loyalty platforms
  • Campaign automation
  • API integrations
  • Journey orchestration
  • Segmentation
  • A/B testing & experimentation
  • End-to-end campaign lifecycle

AI & automation

  • Claude
  • Gemini
  • Groq
  • RAG
  • Prompt engineering
  • AI safety
  • LLM evaluation

SQL & tools

  • Jira
  • Postman
  • Git
  • Docker
  • Grafana
  • Zendesk
  • Figma
02 / Impact delivered

Numbers that explain a business outcome.

Sailer, the portfolio assistant

Sailer · executive summary

“From an executive perspective, these aren’t support metrics — they’re product efficiency gains. Suraj permanently resolved 4,800+ recurring issues and debugged 500+ enterprise campaigns and automations, saving roughly 75+ engineering hours per month in manual triage, while governing 20M+ customer records to enterprise SLA standards. Then he productized what he learned into an AI platform.”

4,800+

Issues resolved permanently

500+

Campaigns & automations debugged and fixed

20M+

Customer records governed

1,500+

Data investigations run

20+

Dashboards and microsites built

95%+

Enterprise SLA sustained

12+

Platform improvements influenced

1

AI platform built end-to-end

WHERE THE FOUR YEARS WENT
4 classes

EXPLAIN MOST OF THE QUEUE

  • Data / integration failures38%
  • Campaign & journey issues27%
  • Loyalty and offer logic21%
  • Everything else14%

Share of incident volume by failure class — four classes explain most of the queue. The 500+ campaign fixes live inside that 27%: enough repetitions to know every failure mode by heart.

OUTCOME SCOREBOARD
Enterprise SLA sustained95%+
Issues closed at root cause82%
Repeat incidents removed64%
Escalations avoided at L147%

Delivery against the enterprise targets I owned as Tech Support Lead, across 15+ enterprise accounts in BFSI, manufacturing, retail and consumer brands.

Recognized for the work — and the way I work.

  • 2× Valuable Employee Award
  • Team Excellence Award
  • Most Consistent Employee
  • Most Respectful FirstHiver
  • 11 recognition awards across 3.5 years at FirstHive

Also competitive off the clock — 2× Best Bowler, Best Batsman, and Player of the Tournament in corporate cricket, a Best Team award, and a badminton doubles title. Consistency isn’t just a work trait.

03 / Flagship case study

AI Marine

GenAI that turns enterprise support incidents into diagnosed, evidence-backed product signal.

After personally debugging 500+ campaign and automation failures, I knew exactly which patterns consumed the support team’s time. In February 2026, I got the mandate to fix it — so I architected, built, and security-tested AI Marine end-to-end.

WHAT AI MARINE CHANGES

Time to understand and create a ticket

BEFORE~30 min
AFTER~30 sec

Tier that can own the first response

BEFOREL2
AFTERL1

Evidence gathered per incident

BEFOREmanual
AFTERcited

Measured against historical incidents with known root cause; currently running in pre-production ahead of full rollout.

Problem

Four failure classes consumed most L2 time.

  • Schema mismatches, segment drift, delivery failures, loyalty gaps
  • ~30 minutes of manual reading and triage before a ticket even existed
  • Skilled engineers spending time on retrieval, not judgement

Architecture

A production-grade stack, not a demo.

  • Django (tickets, webhooks, admin) + PostgreSQL + Redis for caching and rate-limiting
  • React frontend; Microsoft Graph API webhooks for email ingestion with IMAP fallback and idempotent dedup
  • LLM gateway abstraction fronting Claude (primary) and Groq (fallback), with KB/RAG grounding and a cache layer
  • Docker → GKE (Kubernetes CronJobs/Deployments), GitLab CI/CD, AWS Secrets Manager for config

Safety & testing

Assist-first, human-accountable, fully traceable.

  • No customer PII enters a model prompt; no autonomous writes to production systems
  • Every answer cites the artefact it came from
  • End-to-end Playwright test suite covering functionality plus security — XSS, CSRF, prompt injection, rate limiting — mapped to OWASP Top 10 and OWASP LLM Top 10

Impact

Compresses the first thirty minutes to thirty seconds.

  • Piloted in pre-production against historical incidents with known root cause
  • Shifts first-response ownership from L2 down to L1
  • Recurring failure classes surface as prioritised backlog candidates instead of repeat tickets

Built solo — architecture, backend, frontend, infra, and security testing — using AI-assisted development, with every design decision, validation rule, and deployment owned by me.

Technical architecture, pipeline and safety model follow below ↓

04 / Product case studies

Seven investigations, one line each.

Problem, approach, impact. SQL on demand.

Reducing campaign failures through root-cause analysis

Root cause
PROBLEM
Campaign failures were being handled one at a time — I personally debugged and fixed 500+ of them, which is exactly how the recurring patterns became impossible to ignore.
APPROACH
Pre-send validation on the dominant causes, plus a recurring report that made failure-reason distribution visible to product rather than buried in tickets.
IMPACT
Failure handling moved from reactive per-ticket work to a prioritised backlog, and repeat causes stopped re-appearing in the queue every month.
Full case study →

Improving WhatsApp delivery reliability

Data pipeline
PROBLEM
Delivery rates were inconsistent across accounts, and the failures surfaced to the customer as 'the message never arrived' with no further detail.
APPROACH
Clearer failure surfacing back to the campaign UI, template-state checks ahead of send, and documented throughput guidance for high-volume accounts.
IMPACT
Customers could self-diagnose the common cases, and the recurring escalations tied to template state largely disappeared.
Full case study →

Enterprise CDP incident intelligence

SQL
PROBLEM
Incident data existed but was never aggregated.
APPROACH
A standing incident-intelligence report circulated to product and engineering, with the top recurring categories named explicitly each cycle.
IMPACT
Fed 15–20 prioritisation decisions and product fixes over the year, sourced from support data rather than roadmap intuition.
Full case study →

Data quality investigations

Product strategy
PROBLEM
Downstream symptoms — wrong segment sizes, personalisation that misfired, loyalty balances that disagreed with the client's own system — all traced back to ingest quality, but were reported as separate bugs.
APPROACH
Stricter ingest validation on the worst-offending paths, plus visibility into rejects so clients could see what was being dropped and why.
IMPACT
A whole class of 'the data is wrong' escalations became self-explaining, and identity resolution accuracy improved on the affected accounts.
Full case study →

Customer journey analytics

Analytics
PROBLEM
Clients would ask why a journey 'wasn't working' with no shared definition of working, and no per-node visibility into where users actually exited.
APPROACH
Node-level drop-off analysis delivered per account, and a repeatable query pattern the team could run for any journey.
IMPACT
Conversations with clients shifted from 'the platform isn't working' to 'this branch condition is too narrow' — a far more productive place to start.
Full case study →

Support automation

Platform
PROBLEM
Skilled engineering time was being spent on retrieval, not on judgement.
APPROACH
Runbook standardisation, self-serve answers for the top repeat requests, and ultimately the assist-first model behind AI Marine.
IMPACT
Escalation mix shifted toward genuinely novel problems, which is what an L2 team should actually be spending its time on.
Full case study →

Analytics dashboard improvements

Operations
PROBLEM
Dashboards answered questions nobody was asking.
APPROACH
Rebuilt reporting around the questions people were manually recreating, with the recurring cuts scheduled rather than requested.
IMPACT
Ad-hoc report requests dropped and the reporting became the source people quoted in meetings instead of their own spreadsheets.
Full case study →
Sailer, the portfolio assistant

Looking to hire a Product Analyst who has already done the job without the title? Ask Sailer about the 500+ campaigns behind AI Marine, or scroll down to see the career evolution.

05 / My story

Why product.

Support was never ticket resolution. It was continuous customer discovery.

  1. 2019 — 2021

    Operations

    Learned how a business actually runs under load: data, dashboards, networks, servers.

  2. 2022 — 2023

    Rebuild

    Relocated home during covid and used the reset for an Executive PG in Software Development & Cyber Security at IIIT Bengaluru — while joining FirstHive as an L2 Support Engineer.

  3. 2023 — 2024

    Engineering → Leadership

    Promoted twice on performance. Led and trained a team of 6, holding 95%+ SLA across enterprise accounts.

  4. 2024 — now

    Building upstream

    Same loop, before the incident exists: 500+ campaigns debugged, patterns productized into AI Marine.

06 / My philosophy

Every support ticket is customer research in disguise.

My work begins where monitoring ends. I investigate why customers struggle, trace failures across systems and data, identify recurring patterns, and collaborate with engineering to remove the root cause—not just the symptom.

I believe the best Product Analysts are not defined by their title but by their ability to understand customers, interpret data, and drive meaningful product improvements.

Every tool in my stack — Grafana, Elasticsearch, New Relic, AWS CloudWatch, Snowflake, MongoDB, Power BI, Python — was learned under pressure, to solve a real client problem that couldn’t wait. That’s how I’ll learn whatever the next role needs, too.

“I don’t measure success by the number of tickets I close. I measure it by the number of problems that never happen again.”

PERSONAL BRAND STATEMENT

07 / Career

Evolution into product.

Each stage answers one question: what did this role teach me about building better products?

SCOPE OF OWNERSHIP OVER TIME
20192022202320242026
  • 2019 — Operations — one process, one team.
  • 2022 — L2 Support Engineer — failures traced across systems.
  • 2023 — L2 Engineer — promoted on performance; deeper into code and data.
  • 2024 — Tech Support Lead — team of 6 trained and led; 15+ enterprise accounts at 95%+ SLA.
  • 2026 — Product work — AI Marine built end-to-end; decisions made before the incident exists.

Relative scope: systems touched, accounts owned and decisions influenced at each stage.

  1. STAGE 01

    Operations Analyst

    BUSINESS OPERATIONS (2019–2021)

    How a business actually runs underneath the org chart — where cost sits, which processes break under volume, and why the elegant solution loses to the one operations can execute on Monday.

  2. STAGE 02

    L2 Support Engineer → L2 Engineer

    SYSTEM THINKING (2022–2024)

    How systems behave rather than how they are documented. Promoted on performance while completing an Executive PG at IIIT Bengaluru. Reading a failure back to its origin trains you to distrust the first plausible explanation — the single most transferable product skill I have.

  3. STAGE 03

    Tech Support Lead

    CUSTOMER LEADERSHIP (2024–NOW)

    How customers actually experience a product, in their words, under pressure. Led and trained a team of 6, held 95%+ SLA across 15+ enterprise accounts in BFSI, manufacturing, retail and consumer brands, and translated between engineering, business and the account — which is most of the product job.

  4. STAGE 04

    Product Analyst (in practice)

    NEXT CHAPTER

    The same loop, upstream. I've already done the work — 500+ campaigns debugged, A/B tests run, an AI platform architected and shipped solo. The next role makes the title match.

Sailer, the portfolio assistant

“Thank you for reviewing my work. I specialize in turning operational friction into scalable product clarity. If you’re evaluating candidates for a similar role, I am available for an interview to discuss how I can apply these systems-thinking approaches to your platform.”

Sailer, the portfolio assistant mascot

Meet Sailer

Curious about what Suraj builds?

I know the case studies, the stack behind AI Marine, and how to reach him.

Just browsing
Ask Sailer