Problem How it works Capabilities Founder Trust & Data Book a Demo

Know the gap.
Close it faster.

Optimly maps what your workforce can actually do, ranks every skills gap by business impact, and builds the learning path to close it — from the content you already license.

optimly · workspace / cloud-practice Running
VP

We committed to a client capability audit on production Kubernetes in 11 weeks. Cloud Practice is 82 people. Can we staff it internally, or do I need to open requisitions this week?

@skill-radar
Profiled 82 engineers from Darwinbox, Jira labels, GitHub activity and LMS history. 11 at L3+ on production Kubernetes — leadership estimate was 30.
6.2s
@gap-architect
Scored the gap against the audit deadline and client contract value. Impact 96 · Strategic. 24 engineers sit within two dependency steps of the benchmark.
4.8s
@pathmaker
Sequenced 24 individual paths — Linux → containers → Docker → K8s networking → production — skipping steps each engineer already holds. 71% sourced from your Pluralsight licence.
9.1s
@impact-model
Modelled hire vs. train at loaded cost. Train 24, hire 0 — ₹1.9Cr avoided, capability 25 days sooner than a search-and-ramp cycle.
3.4s

Staffed internally. 24 engineers projected to reach L3+ by 14 March — 9 days clear of the audit. Manager approvals queued; nudges scheduled Tue 10:30 and Thu 15:00.

Ask about a team, a role, or a business commitment…
The Problem

Companies spend ₹82,000 per employee a year on training, and 70% of that workforce says none of it applies to their job. The content was never the failure — nobody knows who needs what.

How It Works

Three stages,
running continuously.

Optimly connects to the HRMS, LMS and work systems you already run. No migration, no parallel data entry, no six-week audit. The first org-wide skills map lands in week three.

See it on your own data
skill-radar · ingestSyncing
09:41darwinbox · 612 worker records
09:42jira · 14,880 tickets · 22mo window
09:44github · metadata only, no source
09:47cornerstone · 4,204 completions
09:51574 profiles built · 94% coverage

Map what people can actually do

Proficiency scored 0–5 across an 800-skill taxonomy, inferred from real work signal rather than a form nobody fills in honestly. Every score carries a confidence level, and low-confidence scores go to the manager instead of the dashboard.

gap-architect · rankedQ1 2025
MLOps · strategic96
Kubernetes prod · critical91
Data visualisation87
Spark at scale71
single points of failure14

Rank the gaps by what they cost you

Every gap gets an Impact Score built from business criticality, blast radius, deadline pressure and cost to close. The formula is deterministic and visible — because "the model decided" does not survive a conversation with your CFO.

pathmaker · p.menonOn pace
statistics refresher · 4h
intro to ML · coursera · 6h
supervised learning · 3.2h left
4neural nets · generated module
5model review · internal LMS
45%14 Mar

Close them, person by person

Paths are dependency-sequenced and time-budgeted to the hours someone genuinely has. Steps they've already outgrown get skipped. If they run ahead or fall behind, the path recalculates and the projected close date moves honestly.

What We Believe

Learning content is a solved market. What was never solved is knowing who should be sitting in which course.

Capabilities

Built for the questions
L&D gets asked.

Not a dashboard you check quarterly. A working layer your L&D team, your engineering leads and your CHRO each open for different reasons — and get an answer they can defend.

How we handle the data
org fingerprint · 612 staffQ4
java / springL3.8 · 84 people
sql & modellingL3.6 · 122 people
cloud architectureL2.1 · 34 · thin
mlopsL0.6 · 3 · absent
strategy coverage34% → 61% proj.

Org capability fingerprint

One view of where the company is genuinely deep, where it's thin against peers, and which capabilities don't exist at all. Usually the first document that makes skills a board conversation instead of an HR one.

copilotAnswered
youwho can cover billing if karan leaves?
optnobody at L3+. nearest is r.kulkarni
 at L2 — 3 modules, ~14h to close
 flagged SPOF #2 · criticality 91
youqueue the path for her
optdrafted · awaiting manager approval

Ask it in plain language

Key-person risk, bench depth for a project you haven't staffed yet, whether a team can absorb a new mandate. The copilot answers from your live skills graph and shows which records it used.

decay monitor187 alerts
!!34 certifications already lapsed
!62 expiring within 90 days
·salesforce admin · unused 14mo
·k8s 1.24 track · standard moved
closed68%

Decay and compliance monitoring

Skills atrophy on a measurable curve. Optimly tracks practice recency against each skill's half-life and surfaces lapsed certifications, expiring credentials and capabilities the market has quietly moved past.

What You Get

Four positions we
won't trade away.

01

Targeting, not content

We will never sell you a course library. There is no shortage of good material and no advantage in adding more to a catalogue nobody browses.

02

Capability, not attendance

Completion rate is the metric that's easy to collect and close to meaningless. We measure whether someone can now do a thing they previously couldn't.

03

The employee owns the profile

Every person sees their own record, can dispute any score, and can opt out of inference entirely. Skills data never touches compensation or discipline.

04

Integrate, never migrate

Every migration we ask for is a reason the project quietly dies in month two. We read from Darwinbox, Keka, Cornerstone and the rest, and write back.

Founder
Bhavya Ahuja, founder of Optimly
Mann Agarwal
Founder & CEO
CSE undergrad & researcher · IIT delhi

A retrieval problem wearing an HR costume.

Mann Agarwal is an APM at Flipkart and an IIT Delhi '26 graduate, where his work on ranking and targeting systems led him to the same problem inside corporate L&D: companies own the content but have no way to know who needs what. Optimly is the targeting layer.

Optimly started when that research question collided with a mundane one. Companies hold every input needed to target training well — role definitions in the HRMS, competency ratings in performance reviews, real work signal in Jira and GitHub, thousands of licensed course-hours nobody opens. None of it is connected. So training gets assigned by department, because department is the only signal anyone has. Underneath the HR framing, it's an entity-resolution and dependency-graph problem, and it became tractable roughly the moment language models got good enough to read a job description properly.

"Every CHRO knows their people need upskilling. Almost none can tell you which employee needs which skill by when. That gap is the entire company."

He is deliberate about what Optimly doesn't do: no content hosting, no LMS replacement, and no causal ROI claim the methodology can't defend. Gap scoring is a visible formula rather than a model output, precisely because the number has to survive a conversation with someone's CFO.

Trust & Data

Skills data is career data.

Personally identifiable data is separated from the skills graph at ingestion, and skill records reference pseudonymous identifiers. Customer data is never used to train models shared across tenants. Benchmarking is opt-in and works on aggregates only — never row-level records.

Our engineering integrations read metadata, not content: commit frequency, repository languages and review participation from GitHub; labels, components and epics from Jira; busy-free density from calendars. Source code is never ingested. Ticket bodies are never stored. Calendar event titles and attendees are never read.

SOC 2 Type II · in progress GDPR India DPDP 2023 AWS Mumbai residency Tenant isolation RBAC + SCIM SAML 2.0 SSO AES-256 at rest Zero-retention inference No source code ingestion Employee dispute & opt-out Full export on request
Isolated per tenant
Get Started

Your next skills gap is
forming right now.

Thirty minutes, no deck. We run a live gap analysis on three of your real job descriptions — nothing to set up on your side.