Founders
Twelve years writing these records.
Ten years building the systems that read them.
Two founders, two questions. Neither of us overrules the other inside the other’s question.
A product of Brainy Neurals, an AI engineering company operating since 2023. The registry entry is further down this page.
Domain
She has prepared, issued and reviewed the exact document this product drafts.
“The failure was almost never the science. It was something carried across from the master formula that nobody caught until the record was already on the floor.”
Kinjal Patel · Co-founder and CEO, PharmaSynapse
- 12 years pharmaceutical QA
- BMR and BPR preparation and review
- Validation
- QMS and CAPA
- Quality risk management
- IPQC
- Oral solids and injectables
Since 2025
Independent quality and compliance practice
QMS design and remediation / Validation execution / GMP gap assessment / Documentation systems for manufacturers preparing for inspection
2016 – 2025
Assistant Manager, Quality Assurance · Unison Pharmaceuticals
Process and cleaning validation / Quality risk assessment / Deviation, change control, incident, CAPA, OOS / Site Master File, Validation Master Plan, quality manual / Led regulatory audits and wrote the CAPA responses that closed them
2014 – 2016
Officer, Quality Assurance · Aculife Healthcare and Mediwin Pharmaceuticals
Prepared, issued and revised BMRs and BPRs / Final review of executed records / Annual product quality review / SOP development / IPQC on tablets, capsules, liquids
2014
M.Pharm, Quality Assurance · Gujarat Technological University
B.Pharm, 2012
Every field on a PharmaSynapse record is there because she put it there.
Craft
Ten years putting models into somebody else’s workflow.
“Before this is a pharmaceutical problem it is a retrieval and document-generation problem. Read an approved master formula, produce an issuance-ready record, make every value traceable to its source. ”
Mitesh Patel · Co-founder and CTO, PharmaSynapse
- 10 years AI engineering
- Retrieval-augmented generation
- Document AI
- Healthcare AI
- Regulated-data environments
- Deterministic architectures
Since 2023
Founder and Director · Brainy Neurals
Architecture and commercial ownership / Vision work through to structured document systems
2022 – 2023
Making systems meet a constraint · Quantiphi
C++ point clouds, Intel RealSense and Ouster / TensorRT and ONNX / Hybrid GPU and CPU inference
2015 – 2022
Vision on constrained hardware · eInfochips
C++ on NVIDIA Jetson / Detection, occupancy, wrong-way and anomaly / Model port to Qualcomm and Yocto firmware integration
What transfers
- Clinical-trial eCRF automation, where case report forms are populated from source documents
- Medical report extraction from scanned PDFs
- Structured extraction across research papers, tenders and contracts
- The constraint all three share: a generated statement is worth nothing unless the reader can see where it came from
What does not
- None of it is a GMP batch record
- None was built inside a pharmaceutical manufacturing operation
- He has never held a GMP quality role
- The engineering transfers. The domain does not, and he does not claim it
M.Tech, Embedded Systems, Ganpat University · NVIDIA Certified Associate, AI in the Data Center
How decisions get made here
One worked example.
Proposed
Auto-calculate potency-adjusted quantities
The system already holds the assay value and the formula. The arithmetic is trivial.
01Removed
Before a line of it was written
Potency stays manual. The system pre-prints the formula and leaves the field for a person at issuance.
02Because
It is one of the few points where a person takes responsibility for a number
Automating it removes the check the step exists to create, and removes it invisibly.
03One of us knows what is buildable. The other knows what gets defended in an inspection. Where they disagree, the constraint wins.
Behind the two of us
Two founders is the decision layer, not the whole company.
The engineering capacity was already here. The pharmaceutical judgement is deliberately narrower.
01
Engineering
Six engineers at Brainy Neurals, on production systems across four industries. Not assembled for this product, and not paid for by it.
02
Pharmaceutical judgement
Kinjal’s, and not delegated. Every regulatory statement, every generated field and every stage document carries her sign-off.
03
Format and terminology review
Read by practising quality and production people first. A granulation record does not use the same words at every plant.
NOT
No compliance department
One accountable person with twelve years of practice beats a committee assembled to look like one.
The company
The part a supplier qualification actually asks about.
Not the product. Whether the supplier exists, how long it has, and what happens to your records if it stops.
What you get, and what is not there yet
Three answers, and one list of what is missing.
Your data
Single-tenant. Your instance and your data stay in your environment.
The output
A document in your own format. Not a proprietary file that needs our software to be read.
If we fail
Brainy Neurals has customers and revenue that do not depend on whether PharmaSynapse succeeds.
Not yet
No source-code escrow arrangement. No quality agreement template. Both are on the list before the first deployment.
Running supplier qualification? Say so in the early-access form and send the questionnaire. We answer what can be answered today and mark the rest as not yet, in writing.
Early access
We are letting a small number of manufacturers in early.
If your batch records are prepared by retyping an approved master formula, you are exactly who we want to hear from.
- You send one approved master formula and your current record format
- We draft it, then walk the output line by line
- Kinjal on the record. Mitesh on the system.
There is no pilot yet and no deployment data. What exists is a working system and twelve years of practice behind every field on the record.