MS

Muhammad Sarfraz

Senior ML/AI engineer

Lahore, Punjab, Pakistan

About

web sources + resume
SUMMARY
Lahore-based senior ML/AI engineer whose employment timeline is Systems Limited (2018–2020) → Synaptik (2020–2023) → ZinwaTech (Jan 2024–present); LinkedIn and an external talent profile align on role scope and a COMSATS MS (2017). Current employer context is verifiable at the company level via ZinwaTech’s live site with Lahore address and contact. Prior employer Synaptik shows Lahore contact details and named leadership (no person page for the subject). No uniquely attributable developer handle or IP filings surfaced; multiple identity collisions exist. No person‑named press/talks located; sanctions checks show no match. pk.linkedin.com 1self-evalreval.site 2zinwatech.com 3zinwatech.com 4synaptik.ai 5synaptik.ai 6huggingface.co 7tind.wipo.int 8patents.justia.com 9patents.justia.com 10home.treasury.gov 11un.org 12medium proof

Synthesised from web sources.

CURRENT ROLE
Senior Machine Learning Engineer at ZinwaTech (Jan 2024–present). Role scope (LLMs, RAG, multimodal, vector search, FastAPI microservices, cloud MLOps) is self‑reported on LinkedIn; company presence is corroborated by ZinwaTech’s website with a Lahore address and contact. pk.linkedin.com 1self-evalzinwatech.com 3zinwatech.com 4medium proof

Web sources + resume claims.

40%
Confirmed
21 cited facts
3
Roles
Career arc entries
0
Awards
Formal honours
0
Press features
Major-outlet coverage
0
Quotes
Verbatim from interviews
147
Sources
82 distinct domains

Report

professional-data verification
Collected
2026-05-27
Purpose
professional-data verification
Approx. age
not estimated
Location
Lahore, Punjab, Pakistan
Total experience
≈ 8+ years
Languages
not listed

Hiring assessment

6 role-fits · 3 risks

LLM judgment on best-fit roles, bounded by available evidence.

BEST FIT
Senior Machine Learning Engineer

Also a strong fit

LLM EngineerMLOps EngineerApplied ML EngineerNLP EngineerAI Platform Engineer

Experience spans production LLM/RAG systems, vector databases (FAISS/Qdrant), FastAPI microservices, and cloud MLOps across AWS/Azure/GCP, plus scalable inference on Docker/K8s and Kafka ETL. This maps well to senior IC roles in applied ML and ML platform engineering.

Risks & gaps

Public footprint is sparse: no uniquely attributable developer handle or open-source repository; limited third‑party corroboration beyond company websites and a talent aggregator.
Identity-collision risk is high due to a common name; careful verification is needed to avoid conflating with unrelated academics/inventors.
No press, talks, or meetup listings found to independently surface the subject in Lahore AI/ML circles.
medium proof meetup.com 15

Work Experience

5 confirmations

Web sources + resume claims.

Z
ZinwaTechVerified
Senior Machine Learning Engineer
Jan 2024 – Present
Builds production LLM and RAG systems with vector DBs (FAISS/Qdrant), serves via FastAPI microservices, and operates on AWS/Azure/GCP per self-report.
DC1 colleague confirmed
S
SynaptikVerified
Senior AI/ML Engineer
May 2020 – Dec 2023
Delivered LLM pipelines, recommenders and scalable inference on Docker/K8s; built Kafka ETL and search using Elasticsearch/Qdrant per self-report.
EYCWJDAT4 colleagues confirmed
S
Systems LimitedNot verified
AI Engineer
Jan 2018 – Apr 2020
Built NLP/CV systems and ML microservices; supported CI/CD and data pipelines per self-report.
No colleagues have confirmed yet
pk.linkedin.com 1self-eval reval.site 2 medium proof

What stood out

5 findings · surprise score 7/15

Non-obvious facts that aren't on the subject's LinkedIn profile, ranked by surprise.

other
Company-level address/contact verification supports current Lahore employment context (ZinwaTech site lists a Lahore address and contact).
Direct employer-site contact blocks are a concrete corroborator even when no person page exists for the subject.
other
No uniquely attributable GitHub/Hugging Face/PyPI handle located; collisions like Hugging Face “hsarfraz” do not match this identity.
Many senior ML engineers maintain public repos; absence of a clear handle raises diligence follow‑ups.
🔗
unexpected connection
Identity collisions are frequent: unrelated WIPO editor credit and multiple USPTO patents for a different ‘Muhammad Sarfraz Khaliq.’
Collision risk can mislead diligence unless checked carefully against geography and employer context.
other
No person‑named press, conference talks, or meetup listings surfaced tying this identity to Lahore AI/ML events searched.
Active ML practitioners often appear on local rosters; absence suggests a low‑profile public footprint.
other
A namesake appears on SECP’s Registered Intermediaries list (Karachi), likely unrelated to this subject.
Regulatory records can be conflated across common names; helpful to explicitly rule out.

Personal sites

2 sites

Web properties the subject controls — blog, bio domain, project landing page.

Digital channels

1 link

Subject-owned online presence aggregated from LinkedIn, resume, and personal-signal mentions.

Education

1 institution

Web sources + resume claims.

C
COMSATS University Islamabad
Master of Computer Science Computer Science2015 – 2017
COMSATS University Islamabad — Master of Computer Science, Dec 2015–Dec 2017 (self‑reported; mirrored on Reval). pk.linkedin.com 1self-evalreval.site 2medium proof

People the subject works with

2 entries

Co-founders, business partners, advisors, lead investors — outbound network.

Synthesised from web sources.

Beyond the resume

Synthesised from web sources.

  • No uniquely attributable public developer handle found; collisions dominate (e.g., Hugging Face user “hsarfraz” appears unrelated). huggingface.co 7medium proof
  • High identity-collision risk due to common name; unrelated examples include a WIPO/IntechOpen editor credit and separate inventors named Muhammad Sarfraz Khaliq in USPTO patents. tind.wipo.int 8patents.justia.com 9patents.justia.com 10strong proof

Sources

15 cited
  1. 1
    pk.linkedin.com
    third_party
  2. 2
  3. 3
    zinwatech.com
    third_party
  4. 4
    zinwatech.com
    third_party
  5. 5
    synaptik.ai
    third_party
  6. 6
    synaptik.ai
    third_party
  7. 7
    huggingface.co
    third_party
  8. 8
    tind.wipo.int
    third_party
  9. 9
    patents.justia.com
    third_party
  10. 10
    patents.justia.com
    third_party
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15

Additional sources

Anchor-confirmed pages the summary LLM didn't cite inline — surfaced here so pipeline-validated material isn't hidden.

  1. https://www.meetup.com/topics/ai/pk/ [third_party]