Appendix B · Curriculum Vitae & correspondence

About the Author

A formal record of training, employment, and the experiments that did not make it into the main paper.

B.1Author note

Samarth Suthar is a final-year Computer Engineering undergraduate at the Vidush Somany Institute of Technology and Research, Kadi (2022 — 2026), and a contributor at eInfochips' AI-Studio division. His work centres on retrieval-augmented generation, medical computer vision, and the engineering questions that arise when language models are made to ground their outputs in evidence.

He has authored, contributed to, or shipped a renal-pathology classifier (Kidnex), an open-source cited-RAG document-chat (Kneen), a pothole-measurement pipeline at CSIR · CRRI, and the production embedded-test-case RAG inside AI-Studio. He has solved more than three hundred algorithm problems across LeetCode, Codeforces, and GeeksforGeeks, and graduated from NPTEL's Python for Data Science in the top 2% of his cohort.

B.2Experimental record

B.2.1 Employment

Jan 2026 — present · current affiliation

eInfochips, an Arrow Electronics company

AI-Studio division · Generative AI / RAG

Architecting a RAG system for embedded-device test-case generation — ingesting datasheets, errata, and historical bug reports.
Authoring the next-generation retrieval pipeline for AI-Studio's website test-case generator; lifted engineer-accept rate by 34 pp on the internal benchmark.
Productionising RAG primitives (hybrid retrieval, HyDE, reranking) as reusable internal libraries in Python.

Feb 2025 — May 2025 · research internship

CSIR — Central Road Research Institute

Research intern · Computer vision for civil infrastructure

Built a Flask + OpenCV pipeline to measure pothole area from monocular phone imagery, with a reference object for scale.
Achieved 8.7% MAPEon a 120-sample field set; the project brief required <15%.
Internship deliverable detailed in Appendix A.3.

B.2.2 Education

2022 — 2026 · in progress

B.E., Computer Engineering

Vidush Somany Institute of Technology and Research, Kadi · CGPA 7.50 / 10

Coursework relevant to the present work: Operating Systems, DBMS, Distributed Systems, Computer Vision (elective), Statistical Methods.

B.3Skill matrix

**Table 3.** Self-reported competency across the families of work that recur in this portfolio. The five-point scale is intentionally coarse: I would rather under-promise than over-claim.
Family	Tools & techniques	Self-rated
Generative AI	LLM prompting, structured output, function-calling, evals	4 / 5
Retrieval & RAG	pgvector, BM25, RRF, HyDE, reranking, chunking strategy	5 / 5
Deep learning	PyTorch, transfer-learning (ResNet), fine-tuning, calibration	3 / 5
Computer vision	OpenCV, classical pipelines, CT image preprocessing	4 / 5
Backend	FastAPI, Flask, Python, SSE / streaming, Postgres / DBMS	4 / 5
Infra	Docker, Redis, Postgres + pgvector, basic Linux ops	3 / 5
Algorithms	300+ problems on LeetCode / Codeforces / GFG; 5 LC badges	4 / 5

B.4Achievements & awards

300+ algorithm problems solved across LeetCode, Codeforces, and GeeksforGeeks; five achievement badges on LeetCode covering streaks and topic-specific milestones.
NPTEL · Python for Data Science — 84% final mark; top 2% of national cohort (Swayam programme).
CSIR — CRRI internship— brief delivered; pipeline beat the brief's accuracy target by 6+ percentage points.
Open-source maintainer — Kneen (cited-RAG document chat), released under permissive licence.

B.5Correspondence

Recruiters, collaborators, and curious peers are welcome. The fastest channel is email; for project source, the linked GitHub. The fully formatted BibTeX record below is for the meta-paper-inclined.

Email · primary

sutharsamarth16@gmail.com

Location

Gujarat, India · UTC+05:30

Cite this portfolio

Should the reader wish to refer back to this document in writing, the canonical BibTeX entry is below. The copy button copies the entry to clipboard.

@misc{suthar2026portfolio,
  author       = {Suthar, Samarth},
  title        = {Building Retrieval-Augmented Systems at the
                  Edge of Healthcare and Embedded Intelligence},
  howpublished = {Working portfolio (arXiv:2026.0519v3)},
  year         = {2026},
  month        = {may},
  institution  = {eInfochips, An ARROW Company; VSITR Kadi},
  email        = {sutharsamarth16@gmail.com},
  url          = {https://github.com/Samarth0016},
  keywords     = {RAG, LLM, pgvector, ResNet-50, OpenCV},
  note         = {In revision; recruiter comments welcome.},
}

B.6Acknowledgements

Thanks are owed to the AI-Studio team at eInfochips for the latitude to ship the embedded-test-case RAG into a real product, and to the mentors at CSIR — CRRI who tolerated my insistence on doing the pothole work in OpenCV rather than a pre-trained YOLO. Thanks also to the maintainers of pgvector, FastAPI, and the original PyTorchResNet weights — standing on shoulders, as ever. The remaining errors and overclaims in this document are entirely the author's.

B.7Statements

Availability

Open to full-time roles from June 2026; open to internships and contract work immediately. Remote, India-based, or relocation-friendly.

Funding

This portfolio was self-funded. No external sponsor. Coffee was sourced internally.

Competing interests

The author is currently employed by eInfochips (Arrow). All opinions expressed are personal and do not represent the position of his employer.

Data & code availability

Kneen: open source, see GitHub. Kidnex: code on request, weights gated by dataset licence. Pothole work: deliverable property of CSIR — CRRI; technique reproducible from A.3.