Product Manager - Models
About the job
- At least 3 years of experience building artificial intelligence, applied artificial intelligence, machine learning products, or model training pipelines.
- Familiarity with the model lifecycle, including training, fine-tuning, deployment, and monitoring.
- Strong analytical skills and comfort with data-driven decision-making.
- Ability to collaborate across research, engineering, and product teams.
- Define and drive roadmaps for automatic speech recognition, text-to-speech, and evaluation systems.
- Partner with engineering and research teams on data collection, training cycles, and model release planning.
- Set up and own evaluation frameworks and quality metrics covering accuracy, latency, mean opinion score, and robustness.
- Run structured experiments, including offline tests, A/B tests, and human evaluations, to quantify model improvements.
- Ensure scalable deployment and monitoring of models in collaboration with infrastructure and data teams.
- Translate user and business needs into measurable model goals.
- Communicate priorities, timelines, and trade-offs across stakeholders.
- Exposure to speech technologies, including automatic speech recognition and text-to-speech, or evaluation pipelines.
- Experience with quality metrics such as word error rate, character error rate, mean opinion score, or latency tracking.
- Hands-on familiarity with SQL, Python, or machine learning toolkits.
About the company
Sarvam AI builds a full-stack Generative AI platform and research-informed models for enterprise use in India. It combines training of custom large language models and bespoke enterprise models with an enterprise-grade platform for authoring, deployment, and distribution of AI applications. The product works by offering end-to-end tooling: researchers develop and fine-tune models, then enterprises author, deploy, and manage these models within a scalable platform that supports multilingual and diverse Indian business needs. Sarvam AI differentiates itself through its focus on the Indian market, tailoring AI solutions to linguistic diversity and enterprise requirements, and by pairing model development with deployment infrastructure to deliver cost-effective, robust performance for customer deployments. Its goal is to accelerate the adoption of Generative AI in India by making development, deployment, and distribution of AI applications more efficient and affordable for enterprises.
Company Size
51-200
Company Stage
Series B
Total Funding
$275M
Headquarters
Bengaluru, India
Founded
2023
Get referred to Sarvam
See people who can refer or advise you
Simplify's Take
What believers are saying
- June 2026 Series B raised $234 million, valuing Sarvam at $1.5 billion.
- September 2026 Mahindra Finance passed one crore Sarvam-powered multilingual calls.
- September 2026 Vision 2.1 added handwritten Indian-language OCR and cheaper production serving.
What critics are saying
- HCLTech owns 10.5% after leading June 2026 funding, concentrating strategic dependence.
- Saaras V4 and Vision 2.1 remain API-only; closed weights limit developer lock-in.
- If Odisha compute or enterprise demand slips by 2028, Sarvam's sovereign-AI story breaks.
What makes Sarvam unique
- Sarvam owns India-first sovereign AI, spanning models, voice, vision, and enterprise deployment.
- Saaras V4 covers 22 Indian languages plus English in one API model.
- HCLTech picked Sarvam for the Odisha AI park and 5,000-seat delivery center.
Help us improve and share your feedback! Did you find this helpful?
Benefits
- Hybrid Work Options
Company News
SquadStack launches Arth V2 speech model for indian-language calls. SquadStack launched Arth V2, an in-house speech-to-text model for Indian customer calls. The company reported an 8.85% semantic word error rate and 67-millisecond latency on its own benchmark, while making the model available in closed beta. 12 Oct 2026 16:35 IST SquadStack.ai has launched Arth V2, a speech-to-text model designed for Indian-language customer calls, as the company seeks to reduce transcription errors in voice-based artificial intelligence (AI) systems used for sales and customer interactions. The company said Arth V2 recorded a semantic word error rate of 8.85% and a latency of 67 milliseconds on its internal Conversational Streaming Automatic Speech Recognition (ASR) Benchmark, which uses 863 real Indian sales calls recorded over 8 kHz telephone audio. According to SquadStack, the model produced fewer meaning-changing transcription errors than several competing systems, including models from Sarvam, Google, OpenAI and ElevenLabs. These results are based on the company's own benchmark and have not been independently verified. The Noida-based company said Arth V2 has processed more than one crore minutes of live customer calls over two months. The model is deployed within SquadStack's voice AI agents for enterprise customers and is also available in closed beta for organisations seeking to test it on their own call recordings. Benchmark focuses on indian-language conversations. SquadStack said its benchmark was designed to reflect conditions in Indian contact centres, where customers frequently switch between languages, speak over agents and make calls against background noise. More than 80% of the speech minutes in the test dataset are code-mixed, while nearly 90% contain background noise, according to the company. The recordings use narrowband telephone audio, which captures a more limited frequency range than the wideband recordings commonly used in speech-recognition testing. The benchmark measures semantic word error rate, which focuses on transcription mistakes that alter the intended meaning rather than treating every word substitution as equally consequential. SquadStack reported that Arth V2 recorded fewer meaning-changing errors than Sarvam Saaras v4, Google Gemini 3.5 Transcribe, OpenAI gpt-live-transcribe and ElevenLabs Scribe v2 Realtime. However, the release also acknowledged that Soniox v5 recorded a marginally lower error rate of 8.77%, compared with Arth V2's 8.85%. The company said Arth V2 delivered its final transcript within 67 milliseconds of a customer stopping speaking on 80% of turns. It reported that the model was faster than the competing systems in its tests, although the release did not provide a complete set of comparable latency measurements for every model. SquadStack said the benchmark's test audio, reference transcripts and scoring methodology are available on Hugging Face, allowing other developers to evaluate their own speech-recognition systems against the dataset. Company builds in-house Speech Recognition. Arth V2 is the latest version of SquadStack's in-house speech-recognition technology. The company said it trained the model from scratch using more than 60 crore minutes of Indian sales conversations, covering over 10 crore unique speakers. According to SquadStack, Arth currently handles 80% of the more than 50 lakh AI calls the company makes each day. Its chief technology officer, Vikas Gulati, said the company plans to release updated versions of the model every few weeks as it identifies and addresses transcription errors in production. The approach is intended to improve the speech-recognition component of voice AI agents, where an incorrectly transcribed amount, date or customer response can affect the system's subsequent action. Apurv Agrawal, co-founder and chief executive officer of SquadStack, said transcription remains a significant failure point for voice AI agents because downstream language models depend on correctly capturing what a customer says. The company provides voice AI agents for consumer sales and said it works with more than 60 large brands, including Kotak Mahindra Bank, PhonePe, Angel One and IndiaMART. The launch comes as companies deploy conversational AI systems for customer support, sales and other telephone-based interactions. For these applications, performance on real-world telephone audio may differ from results obtained using clean recordings or general-purpose speech benchmarks. Arth V2's broader performance will depend on how it compares with competing models across independently evaluated datasets, languages and use cases beyond SquadStack's own customer calls.
Sashikumar Sreedharan to be President and CBO at Sarvam AI: Reports. Sashikumar Sreedharan has recently stepped down as Managing Director at Google Cloud India. Sashikumar Sreedharan has been roped in as President and Chief Business Officer at Sarvam AI, as per media reports. In this role, Sreedharan is expected to drive the company's business strategy and supporting its growth as it expands its AI offerings in India and global markets. Sreedharan recently stepped down from his role as Managing Director at Google Cloud India, while Amit Kumar, Managing Director of Digital Natives Business at Google Cloud India, has been appointed as interim leader. Karan Bajwa, President, Asia Pacific, Google Cloud, announced the leadership change in a LinkedIn post. Recently, Sarvam AI was in news after it raised $234 million in the first close of its $300 million Series B at a post-money valuation of $1.5 billion. HCLTech and Bessemer Venture Partners invested in the round, with continued support from existing investors Khosla Ventures and Peak XV Partners.
Sarvam AI has appointed Sashikumar Sreedharan, former Google Cloud India managing director, as president and chief business officer. The company aims to leverage his experience to develop Indian AI infrastructure. The appointment follows Sarvam AI's recent $234 million funding round, which valued the company at $1.5 billion. The startup plans to open a US office to attract Indian-origin AI researchers and has added researcher Devendra Chaplot as an adviser. Sarvam's Saaras V4 model currently leads in 22 Indian languages. The company is developing larger models focused on coding and cybersecurity whilst prioritising digital sovereignty for India.
Sarvam AI releases Saaras V4: A speech-to-text model for all 22 Indian languages and global English. September 26, 2026 Sarvam AI has released Saaras V4, the newest generation of its speech recognition model. It covers all 22 scheduled Indian languages plus English, now including global English accents. Sarvam reports state-of-the-art accuracy across all 22 languages. Is it deployable? Yes, through Sarvam's API today, using model="saaras:v4". Weights are not public, and Sarvam's SageMaker self-hosting docs currently cover Saaras v3 only. What is inside Saaras V4. Saaras V4 is an encoder-decoder system. An audio encoder converts the waveform into embeddings that carry phonetic and acoustic detail. A temporal-downsampling adapter then shortens that sequence and projects it into the language model's embedding space. This keeps long recordings inside the decoder's context budget. The decoder is Sarvam-3B, a 3B-parameter hybrid state-space language model trained from scratch in-house. It reads the audio features alongside a text prompt. It then emits the transcript autoregressively, feeding each token back as input for the next. Benchmark results. * English: Sarvam evaluated 7 English datasets. Six come from Hugging Face's Open ASR Leaderboard: AMI, GigaSpeech, LibriSpeech clean, LibriSpeech other, SPGISpeech and VoxPopuli. The seventh is AI4Bharat's Indian-accented Svarah. Scoring follows the leaderboard's normalization code. Saaras V4 posts the lowest average WER among the models Sarvam benchmarked. * Indic: On Vistaar, Sarvam reports results across 10 Indian languages using both WER and LLM-WER. LLM-WER adds a semantic check. It separates real meaning errors from harmless spelling or formatting variants common in Indic scripts. * Noisy audio: On Kathbath Noisy, measured with LLM-WER, Sarvam says Saaras V4's error rate is under half that of Deepgram Nova-3 and GPT-4o Transcribe. The set includes compressed, clipped and background-heavy recordings. * Language ID: On verified IndicVoices utterances, language identification error is 2.9% across the top 10 Indian languages. It is 5.22% across all 22. It is important to note that all numbers above are vendor-reported. Independent reproduction has not been published yet. 5 output modes from 1 model. The same audio can return 5 representations, selected through the mode parameter: * transcribe (default): native script with numbers and dates normalized. * verbatim: every word as spoken, fillers and spoken numbers kept. * codemix: native script, with English words left in English. * translit: the full utterance in Latin script. * translate: an English translation with numbers normalized. Sarvam's argument is simple. Handling these inside the model removes post-processing steps that can compound errors. Keyterm prompting. Keyterm prompting is new in V4 and works only with saaras:v4. You pass a JSON list under keyterms, with up to 50 terms of 64 characters each. Keyterms bias recognition; they do not guarantee output. Use codemix mode when a brand such as PhonePe must stay in Latin script. On IndicContextEval (paper, Interspeech 2026), Saaras V4 reports 16.03% WER in the L5 keyword-prompting setting. Sarvam says that is the lowest score on the benchmark. Streaming, long audio and pricing. * Streaming: WebSocket with partial results and time to first token below 150 ms. * REST: synchronous transcription for clips up to 30 seconds. * Batch: asynchronous jobs up to 2 hours per file, with optional speaker diarization. * SDKs: Python 3.9+ and Node.js 18+, plus LiveKit Agents, Pipecat and Vercel AI SDK integrations. * Price: Sarvam lists speech-to-text at ₹30 per hour for real-time, streaming and batch, and ₹45 per hour with diarization. Saaras v3 stays the default model. V4 uses the same request shape, so switching is a 1-line change. Saaras V4 vs closest competitors. These are the 3 systems Sarvam benchmarked against. Figures come from each vendor's public docs and pricing pages, checked on September 26, 2026. Key takeaways. * Saaras V4 covers all 22 Indian languages plus global English in 1 model. * A 3B hybrid state-space decoder, trained from scratch, sits behind an audio encoder. * Keyterm prompting accepts up to 50 terms and scored 16.03% WER on IndicContextEval L5. * 5 output modes and sub-150 ms streaming TTFT come from the same model. * API-only today at ₹30 per hour; self-hosting docs still cover v3. Need to partner with Marktechpost LLC. for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with Marktechpost LLC. Asif Razzaq is the CEO of Marktechpost AI Media Inc... As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences.
Sarvam launches Vision 2.1 AI model for documents and Indian languages. Sarvam has launched Vision 2.1, a new AI model that can read forms, tables and documents in English and 22 Indian languages. Sarvam has launched the latest version of its Vision AI model, with a focus on reading and understanding documents. The model can extract information from forms and tables and turn it into structured digital data. Vision 2.1 can also read handwritten text in Indian languages. Sarvam said the model scored 87.3 on the olmOCR-Bench and 87.39 on its own Indic benchmark. Better at forms, tables and handwriting. The new model is designed to handle more complicated documents, including tables spread across multiple pages and forms where information has to be picked out from specific fields. It can also recognise handwritten text in Indian languages. This could be useful for converting older or handwritten documents into digital formats. Sarvam said it worked on reducing incorrect responses and inconsistent results seen in the earlier version of the model. Model trained on real and synthetic data. Sarvam said Vision 2.1 was trained using a mix of real-world and synthetic data. The training material included handwritten and printed forms in multiple languages, along with artificially filled-in forms. The company then used supervised fine-tuning and reinforcement learning with verifiable rewards to improve the model's results. Sarvam first launched its Vision model in February 2026. The earlier version focused on reading text from images and documents, understanding tables and producing structured information. Lower cost of running the model. Sarvam has also made changes to the technology used to run Vision 2.1, which it said allows the model to be offered at a lower cost than originally planned. Its API has also been optimised for companies that want to use the model in their own software. Vision 2.1 uses a vision-language model along with additional systems that help it understand the layout of a document and the order in which its contents need to be read. Conclusion. Sarvam Vision 2.1 is aimed at businesses and other users that need to process large amounts of documents, forms and tables. The model adds support for handwritten text in Indian languages and is designed to handle more complex documents than the earlier version. Disclaimer: This blog has been written exclusively for educational purposes. The securities mentioned are only examples and not recommendations. This does not constitute a personal recommendation/investment advice. It does not aim to influence any individual or entity to make investment decisions. Recipients should conduct their own research and assessments to form an independent opinion about investment decisions. Investments in the securities market are subject to market risks, read all the related documents carefully before investing. Take action on today's market news - Complete your Demat Account Opening now. Published on: Sep 25, 2026, 3:10 PM IST Team Angel One Team Angel One is a group of experienced financial writers that deliver insightful articles on the stock market, IPO, economy, personal finance, commodities and related categories. Open free Demat Account! Join its 3.8 Cr+ happy customers 10 Cr+ DOWNLOADS Enjoy ₹0 Account Opening charges. Get the link to download the App