Melissa Rajaram

I lead products in messy environments—specialized work, constrained teams, changing organizations—with no clean separation between setting direction and getting the product built.

My product career has been shaped by expanding scope. At Amplio, I went from building a content development function to leading three products and two development teams as VP Product. Since then, I’ve co-founded Accord and independently built Arc.

↓ three things to know about how I lead products

Leadership

How I lead products

I lead with clear ownership, make the rationale and tradeoffs behind product decisions clear, and create room for collaborators’ expertise to shape the product.

I lead from strategy through delivery

I set direction and boundaries, work through technical and delivery constraints, make the necessary tradeoffs, and stay involved through implementation.

How this looks in practiceLess

Amplio · Portfolio

Leading three products meant balancing different users, technical foundations, and commitments across the portfolio. I worked directly with two development teams to make tradeoffs, sequence the work, and keep product direction connected to delivery.

Amplio →

Accord · Founder build

I turn regulatory requirements and school district needs into product decisions. From there, I stay involved through design, implementation, and delivery.

Accord →

I create the conditions for strong collaboration

I make roles, priorities, and decision-making clear, share the reasoning behind product choices, and create room for collaborators to contribute their expertise.

How I work with collaboratorsLess

Amplio · Department building

I built Learning Programs & Research into a department spanning speech-language, dyslexia, and research, with managers leading distinct areas, clear roles and priorities, and a shared method for turning research into product content. I worked across Product, Engineering, and customer-facing teams so the department’s expertise could shape product development and support customer-facing work.

Amplio →

Amplio · Reestablishing Product

After a major RIF left me as the sole product person, I reestablished Product with a new distributed development team, rebuilding how we planned work, made decisions, documented requirements, and worked through delivery. I also replaced a high-friction intake process with a simpler way for clinical and customer-facing staff to surface problems early and contribute their subject-matter expertise directly to Product.

Amplio →

I turn domain expertise into product decisions

I use research, regulation, and clinical practice to shape product structure, workflows, and decisions.

How expertise shapes the productLess

Amplio · Evidence-based content

The Speech & Language product needed SLP-facing content that clinicians could trust and use in therapy. I created an evidence-based methodology that connected research to instructional approaches, program structure, exercises, and feedback. It became the standard platform-wide and led to Digital Promise’s Research-Based Design certification.

Amplio →

Accord · Policy to product

Texas changed how special-education funding is determined, replacing a placement-based model with one based on individual student need. I translated the state’s new requirements into product rules and review workflows designed to support accurate determinations and surface cases that need additional judgment before submission.

Accord →

Arc · Preserving clinical signal

Clinical language sampling needs to preserve fillers, atypical productions, distorted speech, and other details that general-purpose speech recognition often removes or misses. I designed Arc around that limitation, using AI transcription as a starting point and giving clinicians the tools to find and repair what the AI loses.

Arc →

How I make product decisions

My goal is to make product decisions by understanding the broader responsibility behind a task and matching each part of the problem to the method best suited to it.

Start with the responsibility

I look beyond the individual task to the broader responsibility the user has to fulfill, then shape the product around what that requires.

Match methods to the problem

I separate complex problems into parts, then use deterministic methods, AI, and expert judgment where each is best suited.

Amplio · The caseload

The product was centered on therapy and materials for individual students. But SLPs were responsible for managing an entire caseload: documenting sessions, reviewing progress across students, and scheduling recurring groups. I shifted the product toward those caseload-level responsibilities, which led to simpler documentation, caseload-wide reporting, and a calendar.

Amplio →

Accord · The district

Texas introduced a new way to determine special-education funding, but the state’s tools supported one student at a time. Districts were responsible for completing those determinations across their entire special-education population. I shaped Accord around dividing the work, reviewing results, handling cases that needed more attention, and bringing the outcomes together at the district level.

Accord →

Arc · The sample

Most language-sampling tools become useful only after a transcript exists. But creating an accurate transcript from a recording is itself a substantial part of the clinician’s job. Arc reduces that burden by using AI transcription as a starting point, then giving clinicians tools to review, correct, and complete the transcript against the recording.

Arc →

Amplio · Content at scale

Developing speech-language content at scale with a small team required different production methods for different kinds of content. I developed deterministic generators for rule-bound content such as articulation exercises, and generative-AI workflows for language content that benefited from more open-ended generation. The team could then spend more of its time reviewing, correcting, and improving output instead of creating every item from scratch.

Amplio →

Accord · Document to determination

Funding determinations depend on information buried in a student’s IEP and on a rubric with explicit rules. AI extracts and structures the source document, deterministic logic applies the state rubric, and the user reviews the proposed determination alongside the supporting evidence.

Accord →

Arc · Sample analysis

Language sampling combines tasks that call for different kinds of support. Arc uses AI transcription to create a starting point, deterministic language-processing methods for measures that can be calculated reliably, and clinician judgment to correct the transcript and make interpretations that depend on what was actually said.

Arc →

Selected work

These three chapters trace how my scope expanded: leading products inside an organization, building one from scratch, and taking one on independently.

Director → Head of Product → VP Product · 2021–2026

Amplio

Amplio is where my scope grew from clinical and instructional work into product and portfolio leadership. As VP Product, I ultimately led Speech & Language, Reading Intervention, and eSpark, working directly with two development teams through major organizational change.

How my scope expandedLess

The products

Amplio’s portfolio included Speech & Language for school-based SLPs, Reading Intervention for therapist-led dyslexia instruction, and later eSpark’s adaptive elementary math and reading product. The products served different users and had different technical foundations, but all sat within the broader challenge of supporting specialized instruction at scale.

What I changed

I created the evidence-based methodology that became the standard for Speech & Language content, helped expand the product into new intervention use cases through the Intervention Hub, and restarted development of Reading Intervention after it was put on hold. I also reestablished Product twice after major reductions in force, rebuilding the processes and cross-functional relationships needed to keep development moving when I was the only product person left.

Co-founder & CPO · 2026–

Accord IEP

Accord gave me the chance to take a product from an ambiguous policy opportunity to a working application. I defined the product from scratch and got much closer to the build itself, using current AI tools to prototype, design, test, and implement at the application level.

Accord’s district dashboard: funding determination progress for a demo district

From policy to productLess

The product

Accord imports a student’s existing IEP and uses it to draft a funding determination under Texas’s new special-education model. District staff review the proposed ratings alongside the supporting evidence, save and resume their work, and manage determinations across the district. The same student record also supports service documentation, goal-progress tracking, district oversight, and a parent view.

What I changed

I took Accord from an ambiguous policy opportunity to a working multi-role product. I defined how the new funding model would translate into product rules and district workflows, built the provider, administrator, parent, and funding experiences, and used AI-assisted development tools to prototype, test, and implement much of the application myself.

Speech and language samples · Independent · 2026–

Arc

arcsampling.com ↗

Arc is where I’m bringing those pieces together independently: product leadership, research, design, hands-on building, and speech-language expertise. I identified a problem I’m unusually well equipped to understand and am taking it from clinical need to working product myself.

Arc’s transcription editor: utterances aligned to the audio waveform, with marking controls

Building independentlyLess

The product

Arc is a speech and language sampling platform for speech-language pathologists. It uses AI transcription to create a starting point, gives clinicians tools to correct the transcript against the recording, and uses the completed transcript as the basis for speech and language measures and reporting.

What I changed

I took a longstanding clinical problem I had wanted to solve and turned it into a product I could build independently. I focused first on the upstream transcription burden, then designed Arc to combine AI assistance, deterministic language-processing methods, and clinician judgment rather than relying on a single technical approach. I own the research, product direction, interaction and visual design, requirements, and working application.

A little more about me

I’m looking for my next product leadership role.

In my free time, I’m building Arc because language sampling is an itchy problem and apparently I have weird hobbies. Those hobbies also include arguing with AI about whether a sentence is true or whether the AI is merely pleased with itself for having written it. Looking at you, Claude.

Away from the computer, I’m the caretaker of Zeb, my charismatic beast who I’ve accidentally trained to find dog people on our walks around town.

If you’d like to talk: hello@melissarajaram.com.